GI Pressure Calculation Statistics

library(readxl)
PWH_data <- read_excel("C:/Users/zhaoz03/OneDrive - The Mount Sinai Hospital/MY PROJECTS/Project 11. GI Pressue EVA Study/Dataset/Final spreadsheet_controls and PWH.xlsx", sheet = 2, skip = 1) %>%
  mutate(
    MCASS_total = 
      case_when(
        mcass_total <= 10 & mcass_total >= 0 ~ "Normal/Equivocal",
        mcass_total <= 20 & mcass_total >= 11 ~ "Mild AN",
        mcass_total <= 30 & mcass_total >= 21 ~ "Moderate AN",
        mcass_total > 30 ~ "Severe AN",
        TRUE ~ NA),
    Autonomic_Neuropathy = 
      case_when(
        cass_total >= 3 ~ "Yes",
        cass_total < 3 ~ "No",
        TRUE ~ NA),      
    Prolonged_SBTT = case_when(
        as.numeric(SBTT...34) > 360 ~ "Yes",
        as.numeric(SBTT...34) <= 360 ~ "No",
        TRUE ~ NA)
    ) %>%
  select(ID, Age, Sex, MCASS_total, Autonomic_Neuropathy, SBTT...34, Prolonged_SBTT) 

Control_data_eva <- read.csv("C:/Users/zhaoz03/OneDrive - The Mount Sinai Hospital/MY PROJECTS/Project 9. EVA new study - GI Motility in HIV/Final spreadsheet_controls_Updated_9.3.24.csv", skip = 1, header = TRUE, na.strings = c("", "NA")) %>% 
  filter(!(SITE %in% c("EVA control group", "Male participants from first three batches (historical control)", "Over 65 years of age (historical controls)"))) %>%
  mutate(AGE = as.numeric(AGE)) %>%
  mutate(
    MCASS_total = 
      case_when(
        mcass_total <= 10 & mcass_total >= 0 ~ "Normal/Equivocal",
        mcass_total <= 20 & mcass_total >= 11 ~ "Mild AN",
        mcass_total <= 30 & mcass_total >= 21 ~ "Moderate AN",
        mcass_total > 30 ~ "Severe AN",
        TRUE ~ NA),
    Autonomic_Neuropathy = 
      case_when(
        cass_total >= 3 ~ "Yes",
        cass_total < 3 ~ "No",
        TRUE ~ NA),    
    Prolonged_SBTT = case_when(
        SBTT > 360 ~ "Yes",
        SBTT <= 360 ~ "No",
        TRUE ~ NA)
    ) %>%
  select(SUBJID, AGE, SEX, MCASS_total, Autonomic_Neuropathy, Prolonged_SBTT)  %>%
  rename(Age = AGE, Sex = SEX)

Control_data_jack <- read_excel("C:/Users/zhaoz03/OneDrive - The Mount Sinai Hospital/MY PROJECTS/Project 11. GI Pressue EVA Study/Dataset/Copy of Demographic info_historical controls.xlsx") 
Control_data_jack <- Control_data_jack %>% 
  mutate(
    AGE = as.numeric(AGE),
    combined_ID = paste0(SITEN, SUBJID),
    SEXO = case_when(
      SEXO == 1 ~ "Male",
      SEXO == 2 ~ "Female",
      TRUE ~ NA
    ))%>%
  mutate(
    Prolonged_SBTT = case_when(
        SBTT > 360 ~ "Yes",
        SBTT <= 360 ~ "No",
        TRUE ~ NA)
    ) %>%
  select(combined_ID, AGE, SEXO, Prolonged_SBTT) %>%
  mutate(combined_ID = as.numeric(combined_ID)) %>%
  rename(Age = AGE, Sex = SEXO) %>%
  filter(!is.na(combined_ID))

1 Jack’s Data Analyses

Control_Patients_5min <- read.csv("C:/Users/zhaoz03/OneDrive - The Mount Sinai Hospital/MY PROJECTS/Project 11. GI Pressue EVA Study/Verified Dataset/Control_Patients_5min.csv")


Control_Patients_5min <- Control_Patients_5min %>%
  filter(!is.na(epoch_type)) %>% 
  group_by(combined_ID) %>% 
  mutate(number_of_fast_epochs = sum(epoch_type == "fast epoch"), number_of_medium_epochs = sum(epoch_type == "medium epoch"), number_of_slow_epochs = sum(epoch_type == "slow epoch")) %>%
  ungroup() %>% 
  mutate(
    number_of_total_epochs = rowSums(across(c(number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs))),
    number_of_fast_epochs_frequency = number_of_fast_epochs/total_no_missing_time,
    number_of_medium_epochs_frequency = number_of_medium_epochs/total_no_missing_time,
    number_of_slow_epochs_frequency = number_of_slow_epochs/total_no_missing_time,
    number_of_total_epochs_frequency = number_of_total_epochs/total_no_missing_time)


Control_Patients_5min <- Control_Patients_5min %>% filter(full_window_size == "Yes", !is.na(epoch_type))
Control_Patients_5min <- Control_Patients_5min %>% filter(prefiltration == FALSE)

Control_Patients_5min <- Control_Patients_5min %>% 
  mutate(time_from_swallow_hour = time_from_swallow_minute/60,
         time_from_gastric_emptying_hour = time_from_gastric_emptying_minute/60)

Control_Patients_5min <- Control_Patients_5min %>%
  mutate(time_fraction = time_from_gastric_emptying_hour/total_time)

Control_Patients_5min <- Control_Patients_5min %>%
  rename(sum_of_amplitude = total_pressure_exposure,
        imputed_sum_of_amplitude = imputed_total_pressure_exposure)

Control_Patients_5min <- Control_Patients_5min %>%
  mutate(motility_index = log(sum_of_amplitude * total_number_of_contractions + 1))

1.1 Describe within-subject RMSSD variability for Control Data

Question: Does it seem true that for fast epochs RMSSD is more consistent?

Answer: It seems true that faster epochs RMSSD is more consistent, especially for RMSSD Time.

rmssd_variability_Control_Patients_5min <- Control_Patients_5min %>%
  group_by(combined_ID, epoch_type) %>%
  summarize(
    mean_RMSSD_Time = mean(epoch_RMSSD_Time),
    median_RMSSD_Time = median(epoch_RMSSD_Time),
    sd_RMSSD_Time = sd(epoch_RMSSD_Time),
    mean_RMSSD_Magnitude = mean(epoch_RMSSD_Magnitude),
    median_RMSSD_Magnitude = median(epoch_RMSSD_Magnitude),
    sd_RMSSD_Magnitude = sd(epoch_RMSSD_Magnitude),
    n_epochs = n(),
    .groups = "drop"
  )

library(ggplot2)
library(gridExtra)

p1 <- ggplot(rmssd_variability_Control_Patients_5min, aes(x = epoch_type, y = sd_RMSSD_Time)) +
  geom_boxplot(outlier.shape = NA) +
  geom_jitter(width = 0.2, alpha = 0.5) +
  labs(
    title = "Standard Deviation of RMSSD Time for each ID by Epoch Type",
    x = "Epoch Type",
    y = "SD of RMSSD Time"
  ) +
  ylim(0, 100) +
  theme(
    text = element_text(size = 14),
    axis.text.x = element_text(size = 14),
    axis.text.y = element_text(size = 14),
    axis.title.x = element_text(size = 18),
    axis.title.y = element_text(size = 18),
    plot.title = element_text(size = 16)
  )

p2 <- ggplot(rmssd_variability_Control_Patients_5min, aes(x = epoch_type, y = sd_RMSSD_Magnitude)) +
  geom_boxplot(outlier.shape = NA) +
  geom_jitter(width = 0.2, alpha = 0.5) +
  labs(
    title = "Standard Deviation of RMSSD Amplitude for each ID by Epoch Type",
    x = "Epoch Type",
    y = "SD of RMSSD Magnitude"
  ) +
  ylim(0, 250) +
  theme(
    text = element_text(size = 14),
    axis.text.x = element_text(size = 14),
    axis.text.y = element_text(size = 14),
    axis.title.x = element_text(size = 18),
    axis.title.y = element_text(size = 18),
    plot.title = element_text(size = 16)
  )

p3 <- ggplot(rmssd_variability_Control_Patients_5min, aes(x = epoch_type, y = median_RMSSD_Time)) +
  geom_boxplot(outlier.shape = NA) +
  geom_jitter(width = 0.2, alpha = 0.5) +
  labs(
    title = "Median of RMSSD Time for each ID by Epoch Type",
    x = "Epoch Type",
    y = "Median of RMSSD Time"
  ) +
  ylim(0, 100) +
  theme(
    text = element_text(size = 14),
    axis.text.x = element_text(size = 14),
    axis.text.y = element_text(size = 14),
    axis.title.x = element_text(size = 18),
    axis.title.y = element_text(size = 18),
    plot.title = element_text(size = 16)
  )
p4 <- ggplot(rmssd_variability_Control_Patients_5min, aes(x = epoch_type, y = median_RMSSD_Magnitude)) +
  geom_boxplot(outlier.shape = NA) +
  geom_jitter(width = 0.2, alpha = 0.5) +
  labs(
    title = "Median of RMSSD Amplitude for each ID by Epoch Type",
    x = "Epoch Type",
    y = "Median of RMSSD Magnitude"
  ) +
  ylim(0, 250) +
  theme(
    text = element_text(size = 14),
    axis.text.x = element_text(size = 14),
    axis.text.y = element_text(size = 14),
    axis.title.x = element_text(size = 18),
    axis.title.y = element_text(size = 18),
    plot.title = element_text(size = 16)
  )
grid.arrange(p1, p2, p3, p4, ncol = 2)

mycontrols <- tableby.control(test = TRUE, total = TRUE,
                              numeric.test = "kwt", cat.test = "chisq",
                              numeric.stats=c("N", "Nmiss", "meansd", "medianq1q3", "min", "max"),
                              cat.stats=c("countpct"),
                              stats.labels = list(
    meansd = "Mean (SD)",
    # medianq1q3 = "Median (Q1, Q3)",
    # range = "Min - Max",
    quantile5 = "5th Percentile",
    quantile95 = "95th Percentile",
    Nmiss2 = "Missing"
  ))

tab.test <- tableby(epoch_type ~ mean_RMSSD_Time + median_RMSSD_Time + sd_RMSSD_Time + mean_RMSSD_Magnitude + median_RMSSD_Magnitude + sd_RMSSD_Magnitude, data = rmssd_variability_Control_Patients_5min, control=mycontrols) 

summary(tab.test)

	fast epoch (N=83)	medium epoch (N=87)	slow epoch (N=90)	Total (N=260)	p value
mean_RMSSD_Time					< 0.001
N	83	87	90	260
Mean (SD)	51.261 (4.429)	64.353 (7.868)	79.844 (9.061)	65.536 (13.852)
Median (Q1, Q3)	50.526 (47.764, 53.527)	62.390 (59.435, 67.901)	80.136 (73.471, 84.824)	63.500 (53.541, 76.360)
Min	44.497	52.110	52.200	44.497
Max	63.850	96.330	107.933	107.933
median_RMSSD_Time					< 0.001
N	83	87	90	260
Mean (SD)	50.465 (4.285)	63.418 (7.848)	78.458 (9.098)	64.489 (13.636)
Median (Q1, Q3)	49.280 (47.425, 53.165)	61.950 (58.100, 65.892)	77.975 (71.345, 84.540)	62.400 (53.095, 73.833)
Min	43.860	52.110	52.200	43.860
Max	63.850	96.330	101.360	101.360
sd_RMSSD_Time					< 0.001
N	72	69	81	222
N-Miss	11	18	9	38
Mean (SD)	7.024 (4.858)	9.032 (5.837)	14.478 (8.132)	10.368 (7.238)
Median (Q1, Q3)	6.227 (4.241, 7.857)	7.976 (4.647, 12.864)	12.595 (8.243, 19.333)	8.303 (5.112, 13.737)
Min	0.134	0.000	2.539	0.000
Max	26.168	27.165	40.985	40.985
mean_RMSSD_Magnitude					< 0.001
N	83	87	90	260
Mean (SD)	70.625 (28.163)	59.799 (24.943)	46.434 (26.284)	58.628 (28.166)
Median (Q1, Q3)	67.335 (50.524, 83.942)	53.579 (41.868, 68.798)	37.122 (29.681, 56.516)	52.975 (38.541, 72.248)
Min	20.160	28.345	19.610	19.610
Max	183.573	149.150	180.510	183.573
median_RMSSD_Magnitude					< 0.001
N	83	87	90	260
Mean (SD)	65.556 (24.956)	56.492 (23.552)	43.154 (26.142)	54.769 (26.484)
Median (Q1, Q3)	61.095 (47.792, 75.675)	49.800 (40.300, 63.960)	35.885 (26.372, 49.825)	48.633 (35.900, 65.963)
Min	20.160	28.345	20.100	20.100
Max	157.550	149.150	180.510	180.510
sd_RMSSD_Magnitude					0.004
N	72	69	81	222
N-Miss	11	18	9	38
Mean (SD)	30.410 (30.947)	24.217 (18.916)	22.499 (22.217)	25.599 (24.660)
Median (Q1, Q3)	24.288 (15.183, 34.176)	18.350 (12.501, 28.706)	13.988 (9.279, 25.430)	18.105 (11.634, 30.590)
Min	0.021	0.225	1.344	0.021
Max	230.546	94.753	107.742	230.546

knitr::kable(rmssd_variability_Control_Patients_5min %>%
               group_by(epoch_type) %>%
                summarize(
                  p5_sd_RMSSD_Time = quantile(sd_RMSSD_Time, 0.05, na.rm = TRUE),
                  p5_median_RMSSD_Time = quantile(median_RMSSD_Time, 0.05, na.rm = TRUE),
                  p5_mean_RMSSD_Time = quantile(mean_RMSSD_Time, 0.05, na.rm = TRUE),
                  p95_sd_RMSSD_Time = quantile(sd_RMSSD_Time, 0.95, na.rm = TRUE),
                  p95_median_RMSSD_Time = quantile(median_RMSSD_Time, 0.95, na.rm = TRUE),
                  p95_mean_RMSSD_Time = quantile(mean_RMSSD_Time, 0.95, na.rm = TRUE),
                  p5_sd_RMSSD_Magnitude = quantile(sd_RMSSD_Magnitude, 0.05, na.rm = TRUE),
                  p5_median_RMSSD_Magnitude = quantile(median_RMSSD_Magnitude, 0.05, na.rm = TRUE),
                  p5_mean_RMSSD_Magnitude = quantile(mean_RMSSD_Magnitude, 0.05, na.rm = TRUE),
                  p95_sd_RMSSD_Magnitude = quantile(sd_RMSSD_Magnitude, 0.95, na.rm = TRUE),
                  p95_median_RMSSD_Magnitude = quantile(median_RMSSD_Magnitude, 0.95, na.rm = TRUE),
                  p95_mean_RMSSD_Magnitude = quantile(mean_RMSSD_Magnitude, 0.95, na.rm = TRUE),
                ), 
  format = "html")

epoch_type	p5_sd_RMSSD_Time	p5_median_RMSSD_Time	p5_mean_RMSSD_Time	p95_sd_RMSSD_Time	p95_median_RMSSD_Time	p95_mean_RMSSD_Time	p5_sd_RMSSD_Magnitude	p5_median_RMSSD_Magnitude	p5_mean_RMSSD_Magnitude	p95_sd_RMSSD_Magnitude	p95_median_RMSSD_Magnitude	p95_mean_RMSSD_Magnitude
fast epoch	1.914226	45.3435	45.95040	15.41039	58.4450	59.81925	8.438141	33.011	33.89300	73.84025	112.14500	116.18890
medium epoch	1.370373	55.7390	56.27400	18.83540	78.8380	80.97400	4.124470	33.923	34.58900	61.87619	97.04500	106.19450
slow epoch	4.582274	66.0525	67.42725	29.73492	93.4595	92.70525	3.681226	21.536	23.35412	65.67032	92.99625	93.11925

p1 <- ggplot(Control_Patients_5min %>% mutate(new_ID = paste0(combined_ID, epoch_type)) %>% filter(epoch_type == "fast epoch"), aes(x = factor(new_ID), y = epoch_RMSSD_Time)) +
  geom_boxplot() +
  labs(
    title = "RMSSD Time Distribution by Fast Epoch",
    x = "ID",
    y = "RMSSD Time"
  ) +
  ylim(0, 200) +
  theme_minimal()

p2 <- ggplot(Control_Patients_5min %>% mutate(new_ID = paste0(combined_ID, epoch_type)) %>% filter(epoch_type == "medium epoch"), aes(x = factor(new_ID), y = epoch_RMSSD_Time)) +
  geom_boxplot() +
  labs(
    title = "RMSSD Time Distribution by Medium Epoch",
    x = "ID",
    y = "RMSSD Time"
  ) +
  ylim(0, 200) +
  theme_minimal()

p3 <- ggplot(Control_Patients_5min %>% mutate(new_ID = paste0(combined_ID, epoch_type)) %>% filter(epoch_type == "slow epoch"), aes(x = factor(new_ID), y = epoch_RMSSD_Time)) +
  geom_boxplot() +
  labs(
    title = "RMSSD Time Distribution by Slow Epoch",
    x = "ID",
    y = "RMSSD Time"
  ) +
  ylim(0, 200) +
  theme_minimal()

grid.arrange(p1, p2, p3, ncol = 3)

p1 <- ggplot(Control_Patients_5min %>% mutate(new_ID = paste0(combined_ID, epoch_type)) %>% filter(epoch_type == "fast epoch"), aes(x = factor(new_ID), y = epoch_RMSSD_Magnitude)) +
  geom_boxplot() +
  labs(
    title = "RMSSD Magnitude Distribution by Fast Epoch",
    x = "ID",
    y = "RMSSD Magnitude"
  ) +
  ylim(0, 400) +
  theme_minimal()

p2 <- ggplot(Control_Patients_5min %>% mutate(new_ID = paste0(combined_ID, epoch_type)) %>% filter(epoch_type == "medium epoch"), aes(x = factor(new_ID), y = epoch_RMSSD_Magnitude)) +
  geom_boxplot() +
  labs(
    title = "RMSSD Magnitude Distribution by Medium Epoch",
    x = "ID",
    y = "RMSSD Magnitude"
  ) +
  ylim(0, 400) +
  theme_minimal()

p3 <- ggplot(Control_Patients_5min %>% mutate(new_ID = paste0(combined_ID, epoch_type)) %>% filter(epoch_type == "slow epoch"), aes(x = factor(new_ID), y = epoch_RMSSD_Magnitude)) +
  geom_boxplot() +
  labs(
    title = "RMSSD Magnitude Distribution by Slow Epoch",
    x = "ID",
    y = "RMSSD Magnitude"
  ) +
  ylim(0, 400) +
  theme_minimal()

grid.arrange(p1, p2, p3, ncol = 3)

1.2 Describe how the epochs are distributed over time

Question: Are they more in the proximal or distal small bowel?

p1 <- ggplot(Control_Patients_5min %>% filter(epoch_type == "fast epoch"), aes(x = time_fraction)) +
  geom_histogram(fill = "steelblue", color = "white") +
  labs(
    x = "        ",
    y = "Fast Rhythmic Intervals"
  ) +
  ylim(0, 35) +
  theme(
    text = element_text(size = 14),
    axis.text.x = element_text(size = 14),
    axis.text.y = element_text(size = 14),
    axis.title.x = element_text(size = 18),
    axis.title.y = element_text(size = 18),
    plot.title = element_text(size = 16)
  )

p2 <- ggplot(Control_Patients_5min %>% filter(epoch_type == "medium epoch"), aes(x = time_fraction)) +
  geom_histogram(fill = "steelblue", color = "white") +
  labs(
    x = "Time Fraction since Gastric Emptying Time",
    y = "Medium Rhythmic Intervals"
  ) +
  ylim(0, 35) +
  theme(
    text = element_text(size = 14),
    axis.text.x = element_text(size = 14),
    axis.text.y = element_text(size = 14),
    axis.title.x = element_text(size = 18),
    axis.title.y = element_text(size = 18),
    plot.title = element_text(size = 16)
  )

p3 <- ggplot(Control_Patients_5min %>% filter(epoch_type == "slow epoch"), aes(x = time_fraction)) +
  geom_histogram(fill = "steelblue", color = "white") +
  labs(
    x = "        ",
    y = "Slow  Rhythmic Intervals"
  ) +
  ylim(0, 35) +
  theme(
    text = element_text(size = 14),
    axis.text.x = element_text(size = 14),
    axis.text.y = element_text(size = 14),
    axis.title.x = element_text(size = 18),
    axis.title.y = element_text(size = 18),
    plot.title = element_text(size = 16)
  )

grid.arrange(p1, p2, p3, ncol = 3)

p1

p2

p3

Question: Do they tend to occur together in clumps or separated out?

fast_epochs <- Control_Patients_5min %>% 
  filter(epoch_type == "fast epoch", !is.na(time_from_gastric_emptying_minute)) %>%
  arrange(combined_ID, time_from_gastric_emptying_minute) %>%
  group_by(combined_ID) %>%
  mutate(time_gap = time_from_gastric_emptying_minute - lag(time_from_gastric_emptying_minute) - 5) %>%
  ungroup()

fast_epochs <- fast_epochs %>%
  group_by(combined_ID) %>%
  summarize(
    n_fast_epochs = n(),
    avg_gap = round(mean(time_gap, na.rm = TRUE),2),
    median_gap = round(median(time_gap, na.rm = TRUE),2),
    min_gap = round(min(time_gap, na.rm = TRUE),2),
    max_gap = round(max(time_gap, na.rm = TRUE),2)
  )

medium_epochs <- Control_Patients_5min %>% 
  filter(epoch_type == "medium epoch", !is.na(time_from_gastric_emptying_minute)) %>%
  arrange(combined_ID, time_from_gastric_emptying_minute) %>%
  group_by(combined_ID) %>%
  mutate(time_gap = time_from_gastric_emptying_minute - lag(time_from_gastric_emptying_minute)) %>%
  ungroup()

medium_epochs <- medium_epochs %>%
  group_by(combined_ID) %>%
  summarize(
    n_medium_epochs = n(),
    avg_gap = round(mean(time_gap, na.rm = TRUE),2),
    median_gap = round(median(time_gap, na.rm = TRUE),2),
    min_gap = round(min(time_gap, na.rm = TRUE),2),
    max_gap = round(max(time_gap, na.rm = TRUE),2)
  )

slow_epochs <- Control_Patients_5min %>% 
  filter(epoch_type == "slow epoch", !is.na(time_from_gastric_emptying_minute)) %>%
  arrange(combined_ID, time_from_gastric_emptying_minute) %>%
  group_by(combined_ID) %>%
  mutate(time_gap = time_from_gastric_emptying_minute - lag(time_from_gastric_emptying_minute)) %>%
  ungroup()

slow_epochs <- slow_epochs %>%
  group_by(combined_ID) %>%
  summarize(
    n_slow_epochs = n(),
    avg_gap = round(mean(time_gap, na.rm = TRUE),2),
    median_gap = round(median(time_gap, na.rm = TRUE),2),
    min_gap = round(min(time_gap, na.rm = TRUE),2),
    max_gap = round(max(time_gap, na.rm = TRUE),2)
  )

1.3 Epoch Type - Time Fraction Plot

Control_Patients_5min <- Control_Patients_5min %>% 
  group_by(combined_ID) %>%
  mutate(time_fraction_end = time_fraction + 5/60/total_time)

# ggplot(Control_Patients_5min) +
#   geom_segment(aes(x = time_fraction, xend = time_fraction_end, 
#                    y = as.character(combined_ID), yend = as.character(combined_ID), color = epoch_type)) +
#   labs(x = "Time Fraction", y = "ID", color = "Epoch Type") +
#   theme_minimal() +
#   theme(axis.text.y = element_text(size = 7))

Control_Patients_5min$ID <- as.integer(factor(as.factor(Control_Patients_5min$combined_ID)))
Control_Patients_5min <- Control_Patients_5min %>%
  mutate(ymin = ID - 0.5,
         ymax = ID + 0.5)

ggplot(Control_Patients_5min %>% mutate(
  epoch_type = case_when(
    epoch_type == "fast epoch" ~ "fast",
    epoch_type == "medium epoch" ~ "medium",
    epoch_type == "slow epoch" ~ "slow"
  )
)) +
  geom_rect(aes(xmin = time_fraction, xmax = time_fraction_end,
                ymin = ymin, ymax = ymax, fill = epoch_type)) +
  scale_y_continuous(
    breaks = Control_Patients_5min$ID,
    labels = Control_Patients_5min$combined_ID
  ) +
  labs(x = "Time Fraction since Gastric Emptying Time", y = "Participant ID", fill = "Type of Rhythmic Intervals") +
  theme(
    text = element_text(size = 14),
    axis.text.x = element_text(size = 14),
    axis.text.y = element_blank(),
    axis.title.x = element_text(size = 18),
    axis.title.y = element_text(size = 18),
    plot.title = element_text(size = 16)
  ) +   
  geom_segment(
    aes(x = -Inf, xend = Inf, y = ymin, yend = ymin),
    color = "black", linewidth = 0.02
  )

1.4 Is Total Time Correlated to Number of Epochs?

library(corrplot)
res <- cor(data.frame(
  total_time = (Control_Patients_5min %>% select(combined_ID, total_time, total_no_missing_time, number_of_total_epochs, number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs) %>% unique())$total_time,
  total_no_missing_time = (Control_Patients_5min %>% select(combined_ID, total_time, total_no_missing_time, number_of_total_epochs, number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs) %>% unique())$total_no_missing_time,
  number_of_total_epochs = (Control_Patients_5min %>% select(combined_ID, total_time, total_no_missing_time, number_of_total_epochs, number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs) %>% unique())$number_of_total_epochs,
  number_of_fast_epochs = (Control_Patients_5min %>% select(combined_ID, total_time, total_no_missing_time, number_of_total_epochs, number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs) %>% unique())$number_of_fast_epochs,
  number_of_medium_epochs = (Control_Patients_5min %>% select(combined_ID, total_time, total_no_missing_time, number_of_total_epochs, number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs) %>% unique())$number_of_medium_epochs,
  number_of_slow_epochs = (Control_Patients_5min %>% select(combined_ID, total_time, total_no_missing_time, number_of_total_epochs, number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs) %>% unique())$number_of_slow_epochs
  ), 
  use = "complete.obs", method = "spearman")
# res
corrplot(res, type = "upper", 
         tl.col = "black", tl.srt = 45)

2 EVA Data Analyses

EVA_Patients_5min <- read.csv("C:/Users/zhaoz03/OneDrive - The Mount Sinai Hospital/MY PROJECTS/Project 11. GI Pressue EVA Study/Verified Dataset/EVA_Patients_5min.csv")

EVA_Patients_5min <- EVA_Patients_5min %>%
  filter(!is.na(epoch_type)) %>%
  group_by(combined_ID) %>%
  mutate(number_of_fast_epochs = sum(epoch_type == "fast epoch"), number_of_medium_epochs = sum(epoch_type == "medium epoch"), number_of_slow_epochs = sum(epoch_type == "slow epoch")) %>%
  ungroup() %>%
  mutate(
    number_of_total_epochs = rowSums(across(c(number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs))),
    number_of_fast_epochs_frequency = number_of_fast_epochs/total_no_missing_time,
    number_of_medium_epochs_frequency = number_of_medium_epochs/total_no_missing_time,
    number_of_slow_epochs_frequency = number_of_slow_epochs/total_no_missing_time,
    number_of_total_epochs_frequency = number_of_total_epochs/total_no_missing_time)

EVA_Patients_5min <- EVA_Patients_5min %>% filter(full_window_size == "Yes", !is.na(epoch_type))
EVA_Patients_5min <- EVA_Patients_5min %>% filter(prefiltration == FALSE)

EVA_Patients_5min <- EVA_Patients_5min %>% 
  mutate(time_from_swallow_hour = time_from_swallow_minute/60,
         time_from_gastric_emptying_hour = time_from_gastric_emptying_minute/60)

EVA_Patients_5min <- EVA_Patients_5min %>%
  mutate(time_fraction = time_from_gastric_emptying_hour/total_time)

EVA_Patients_5min <- EVA_Patients_5min %>%
  rename(sum_of_amplitude = total_pressure_exposure,
        imputed_sum_of_amplitude = imputed_total_pressure_exposure)

EVA_Patients_5min <- EVA_Patients_5min %>%
  mutate(motility_index = log(sum_of_amplitude * total_number_of_contractions + 1))

EVA_Patients_5min <- EVA_Patients_5min %>% filter(combined_ID != 7180) # Missing Info

2.1 Describe within-subject RMSSD variability for Control Data

Question: Does it seem true that for fast epochs RMSSD is more consistent?

Answer: It seems true that faster epochs RMSSD is more consistent, especially for RMSSD Time.

rmssd_variability_EVA_Patients_5min <- EVA_Patients_5min %>%
  group_by(combined_ID, epoch_type) %>%
  summarize(
    mean_RMSSD_Time = mean(epoch_RMSSD_Time),
    median_RMSSD_Time = median(epoch_RMSSD_Time),
    sd_RMSSD_Time = sd(epoch_RMSSD_Time),
    mean_RMSSD_Magnitude = mean(epoch_RMSSD_Magnitude),
    median_RMSSD_Magnitude = median(epoch_RMSSD_Magnitude),
    sd_RMSSD_Magnitude = sd(epoch_RMSSD_Magnitude),
    n_epochs = n(),
    .groups = "drop"
  )

p1 <- ggplot(rmssd_variability_EVA_Patients_5min, aes(x = epoch_type, y = sd_RMSSD_Time)) +
  geom_boxplot(outlier.shape = NA) +
  geom_jitter(width = 0.2, alpha = 0.5) +
  labs(
    title = "Standard Deviation of RMSSD Time for each ID by Epoch Type",
    x = "Epoch Type",
    y = "SD of RMSSD Time"
  ) +
  ylim(0, 140) +
  theme(
    text = element_text(size = 14),
    axis.text.x = element_text(size = 14),
    axis.text.y = element_text(size = 14),
    axis.title.x = element_text(size = 18),
    axis.title.y = element_text(size = 18),
    plot.title = element_text(size = 16)
  )

p2 <- ggplot(rmssd_variability_EVA_Patients_5min, aes(x = epoch_type, y = sd_RMSSD_Magnitude)) +
  geom_boxplot(outlier.shape = NA) +
  geom_jitter(width = 0.2, alpha = 0.5) +
  labs(
    title = "Standard Deviation of RMSSD Amplitude for each ID by Epoch Type",
    x = "Epoch Type",
    y = "SD of RMSSD Magnitude"
  ) +
  ylim(0, 200) +
  theme(
    text = element_text(size = 14),
    axis.text.x = element_text(size = 14),
    axis.text.y = element_text(size = 14),
    axis.title.x = element_text(size = 18),
    axis.title.y = element_text(size = 18),
    plot.title = element_text(size = 16)
  )

p3 <- ggplot(rmssd_variability_EVA_Patients_5min, aes(x = epoch_type, y = median_RMSSD_Time)) +
  geom_boxplot(outlier.shape = NA) +
  geom_jitter(width = 0.2, alpha = 0.5) +
  labs(
    title = "Median of RMSSD Time for each ID by Epoch Type",
    x = "Epoch Type",
    y = "Median of RMSSD Time"
  ) +
  ylim(0, 140) +
  theme(
    text = element_text(size = 14),
    axis.text.x = element_text(size = 14),
    axis.text.y = element_text(size = 14),
    axis.title.x = element_text(size = 18),
    axis.title.y = element_text(size = 18),
    plot.title = element_text(size = 16)
  )

p4 <- ggplot(rmssd_variability_EVA_Patients_5min, aes(x = epoch_type, y = median_RMSSD_Magnitude)) +
  geom_boxplot(outlier.shape = NA) +
  geom_jitter(width = 0.2, alpha = 0.5) +
  labs(
    title = "Median of RMSSD Amplitude for each ID by Epoch Type",
    x = "Epoch Type",
    y = "Median of RMSSD Magnitude"
  ) +
  ylim(0, 200) +
  theme(
    text = element_text(size = 14),
    axis.text.x = element_text(size = 14),
    axis.text.y = element_text(size = 14),
    axis.title.x = element_text(size = 18),
    axis.title.y = element_text(size = 18),
    plot.title = element_text(size = 16)
  )

grid.arrange(p1, p2, p3, p4, ncol = 2)

mycontrols <- tableby.control(test = TRUE, total = TRUE,
                              numeric.test = "kwt", cat.test = "chisq",
                              numeric.stats=c("N", "Nmiss", "meansd", "medianq1q3", "min", "max"),
                              cat.stats=c("countpct"),
                              stats.labels = list(
    meansd = "Mean (SD)",
    # medianq1q3 = "Median (Q1, Q3)",
    # range = "Min - Max",
    quantile5 = "5th Percentile",
    quantile95 = "95th Percentile",
    Nmiss2 = "Missing"
  ))

tab.test <- tableby(epoch_type ~ mean_RMSSD_Time + median_RMSSD_Time + sd_RMSSD_Time + mean_RMSSD_Magnitude + median_RMSSD_Magnitude + sd_RMSSD_Magnitude, data = rmssd_variability_EVA_Patients_5min, control=mycontrols) 

summary(tab.test)

	fast epoch (N=61)	medium epoch (N=58)	slow epoch (N=59)	Total (N=178)	p value
mean_RMSSD_Time					< 0.001
N	61	58	59	178
Mean (SD)	54.216 (6.088)	68.275 (10.305)	85.045 (13.828)	69.016 (16.454)
Median (Q1, Q3)	53.943 (49.080, 57.485)	65.690 (62.031, 72.868)	82.040 (76.702, 91.950)	65.590 (55.995, 77.910)
Min	43.885	51.220	64.010	43.885
Max	71.790	101.345	138.230	138.230
median_RMSSD_Time					< 0.001
N	61	58	59	178
Mean (SD)	53.136 (6.481)	67.132 (9.843)	83.876 (14.217)	67.885 (16.494)
Median (Q1, Q3)	52.540 (48.160, 57.180)	65.590 (60.888, 70.885)	80.715 (73.755, 89.062)	65.455 (55.622, 76.742)
Min	41.540	50.480	64.010	41.540
Max	71.790	101.345	138.230	138.230
sd_RMSSD_Time					< 0.001
N	47	40	44	131
N-Miss	14	18	15	47
Mean (SD)	9.304 (5.801)	12.509 (10.274)	15.703 (9.257)	12.432 (8.892)
Median (Q1, Q3)	8.389 (4.755, 12.623)	10.004 (4.393, 17.119)	13.932 (10.789, 18.330)	10.753 (6.347, 16.092)
Min	0.530	1.770	0.559	0.530
Max	25.172	51.456	50.982	51.456
mean_RMSSD_Magnitude					< 0.001
N	61	58	59	178
Mean (SD)	78.622 (29.029)	63.031 (24.903)	51.258 (25.158)	64.472 (28.647)
Median (Q1, Q3)	76.396 (62.313, 90.594)	59.629 (46.562, 72.094)	47.182 (37.784, 62.575)	61.278 (43.933, 79.232)
Min	24.450	27.860	17.460	17.460
Max	158.900	151.905	177.435	177.435
median_RMSSD_Magnitude					< 0.001
N	61	58	59	178
Mean (SD)	73.035 (28.061)	60.919 (25.692)	47.339 (24.965)	60.570 (28.207)
Median (Q1, Q3)	72.370 (54.360, 86.700)	53.445 (42.785, 71.294)	43.950 (34.177, 49.892)	52.252 (40.903, 78.102)
Min	24.450	27.860	17.460	17.460
Max	158.900	151.905	177.435	177.435
sd_RMSSD_Magnitude					0.077
N	47	40	44	131
N-Miss	14	18	15	47
Mean (SD)	33.781 (22.912)	26.299 (22.139)	27.737 (24.361)	29.466 (23.236)
Median (Q1, Q3)	26.073 (17.804, 44.129)	22.894 (11.119, 33.471)	20.582 (11.211, 34.803)	22.947 (13.109, 39.995)
Min	4.778	3.691	1.683	1.683
Max	117.665	129.606	98.889	129.606

knitr::kable(rmssd_variability_EVA_Patients_5min %>%
               group_by(epoch_type) %>%
                summarize(
                  p5_sd_RMSSD_Time = quantile(sd_RMSSD_Time, 0.05, na.rm = TRUE),
                  p5_median_RMSSD_Time = quantile(median_RMSSD_Time, 0.05, na.rm = TRUE),
                  p5_mean_RMSSD_Time = quantile(mean_RMSSD_Time, 0.05, na.rm = TRUE),
                  p95_sd_RMSSD_Time = quantile(sd_RMSSD_Time, 0.95, na.rm = TRUE),
                  p95_median_RMSSD_Time = quantile(median_RMSSD_Time, 0.95, na.rm = TRUE),
                  p95_mean_RMSSD_Time = quantile(mean_RMSSD_Time, 0.95, na.rm = TRUE),
                  p5_sd_RMSSD_Magnitude = quantile(sd_RMSSD_Magnitude, 0.05, na.rm = TRUE),
                  p5_median_RMSSD_Magnitude = quantile(median_RMSSD_Magnitude, 0.05, na.rm = TRUE),
                  p5_mean_RMSSD_Magnitude = quantile(mean_RMSSD_Magnitude, 0.05, na.rm = TRUE),
                  p95_sd_RMSSD_Magnitude = quantile(sd_RMSSD_Magnitude, 0.95, na.rm = TRUE),
                  p95_median_RMSSD_Magnitude = quantile(median_RMSSD_Magnitude, 0.95, na.rm = TRUE),
                  p95_mean_RMSSD_Magnitude = quantile(mean_RMSSD_Magnitude, 0.95, na.rm = TRUE),
                ), 
  format = "html")

epoch_type	p5_sd_RMSSD_Time	p5_median_RMSSD_Time	p5_mean_RMSSD_Time	p95_sd_RMSSD_Time	p95_median_RMSSD_Time	p95_mean_RMSSD_Time	p5_sd_RMSSD_Magnitude	p5_median_RMSSD_Magnitude	p5_mean_RMSSD_Magnitude	p95_sd_RMSSD_Magnitude	p95_median_RMSSD_Magnitude	p95_mean_RMSSD_Magnitude
fast epoch	2.218913	45.1050	46.38000	20.80387	63.55000	64.08714	7.056163	30.3700	31.5700	75.51322	120.1500	130.9700
medium epoch	2.220669	54.7290	56.17683	30.43147	82.58425	89.73275	5.619383	30.2185	30.2185	53.54059	109.6155	103.1257
slow epoch	4.889204	67.1335	69.06625	32.70255	112.02900	112.02900	2.264742	22.3850	20.9756	77.89687	87.6005	83.2505

p1 <- ggplot(EVA_Patients_5min %>% mutate(new_ID = paste0(combined_ID, epoch_type)) %>% filter(epoch_type == "fast epoch"), aes(x = factor(new_ID), y = epoch_RMSSD_Time)) +
  geom_boxplot() +
  labs(
    title = "RMSSD Time Distribution by Fast Epoch",
    x = "ID",
    y = "RMSSD Time"
  ) +
  ylim(0, 200) +
  theme_minimal()

p2 <- ggplot(EVA_Patients_5min %>% mutate(new_ID = paste0(combined_ID, epoch_type)) %>% filter(epoch_type == "medium epoch"), aes(x = factor(new_ID), y = epoch_RMSSD_Time)) +
  geom_boxplot() +
  labs(
    title = "RMSSD Time Distribution by Medium Epoch",
    x = "ID",
    y = "RMSSD Time"
  ) +
  ylim(0, 200) +
  theme_minimal()

p3 <- ggplot(EVA_Patients_5min %>% mutate(new_ID = paste0(combined_ID, epoch_type)) %>% filter(epoch_type == "slow epoch"), aes(x = factor(new_ID), y = epoch_RMSSD_Time)) +
  geom_boxplot() +
  labs(
    title = "RMSSD Time Distribution by Slow Epoch",
    x = "ID",
    y = "RMSSD Time"
  ) +
  ylim(0, 200) +
  theme_minimal()

grid.arrange(p1, p2, p3, ncol = 3)

p1 <- ggplot(EVA_Patients_5min %>% mutate(new_ID = paste0(combined_ID, epoch_type)) %>% filter(epoch_type == "fast epoch"), aes(x = factor(new_ID), y = epoch_RMSSD_Magnitude)) +
  geom_boxplot() +
  labs(
    title = "RMSSD Magnitude Distribution by Fast Epoch",
    x = "ID",
    y = "RMSSD Magnitude"
  ) +
  ylim(0, 400) +
  theme_minimal()

p2 <- ggplot(EVA_Patients_5min %>% mutate(new_ID = paste0(combined_ID, epoch_type)) %>% filter(epoch_type == "medium epoch"), aes(x = factor(new_ID), y = epoch_RMSSD_Magnitude)) +
  geom_boxplot() +
  labs(
    title = "RMSSD Magnitude Distribution by Medium Epoch",
    x = "ID",
    y = "RMSSD Magnitude"
  ) +
  ylim(0, 400) +
  theme_minimal()

p3 <- ggplot(EVA_Patients_5min %>% mutate(new_ID = paste0(combined_ID, epoch_type)) %>% filter(epoch_type == "slow epoch"), aes(x = factor(new_ID), y = epoch_RMSSD_Magnitude)) +
  geom_boxplot() +
  labs(
    title = "RMSSD Magnitude Distribution by Slow Epoch",
    x = "ID",
    y = "RMSSD Magnitude"
  ) +
  ylim(0, 400) +
  theme_minimal()

grid.arrange(p1, p2, p3, ncol = 3)

2.2 Describe how the epochs are distributed over time

Question: Are they more in the proximal or distal small bowel?

p1 <- ggplot(EVA_Patients_5min %>% filter(epoch_type == "fast epoch"), aes(x = time_fraction)) +
  geom_histogram(fill = "steelblue", color = "white") +
  labs(
    x = "        ",
    y = "Fast Rhythmic Intervals"
  ) +
  ylim(0, 35) +
  theme(
    text = element_text(size = 14),
    axis.text.x = element_text(size = 14),
    axis.text.y = element_text(size = 14),
    axis.title.x = element_text(size = 18),
    axis.title.y = element_text(size = 18),
    plot.title = element_text(size = 16)
  )
p2 <- ggplot(EVA_Patients_5min %>% filter(epoch_type == "medium epoch"), aes(x = time_fraction)) +
  geom_histogram(fill = "steelblue", color = "white") +
  labs(
    x = "Time Fraction since Gastric Emptying Time",
    y = "Medium Rhythmic Intervals"
  ) +
  ylim(0, 35) +
  theme(
    text = element_text(size = 14),
    axis.text.x = element_text(size = 14),
    axis.text.y = element_text(size = 14),
    axis.title.x = element_text(size = 18),
    axis.title.y = element_text(size = 18),
    plot.title = element_text(size = 16)
  )
p3 <- ggplot(EVA_Patients_5min %>% filter(epoch_type == "slow epoch"), aes(x = time_fraction)) +
  geom_histogram(fill = "steelblue", color = "white") +
  labs(
    x = "        ",
    y = "Slow Rhythmic Intervals"
  ) +
  ylim(0, 35) +
  theme(
    text = element_text(size = 14),
    axis.text.x = element_text(size = 14),
    axis.text.y = element_text(size = 14),
    axis.title.x = element_text(size = 18),
    axis.title.y = element_text(size = 18),
    plot.title = element_text(size = 16)
  )

grid.arrange(p1, p2, p3, ncol = 3)

p1

p2

p3

Question: Do they tend to occur together in clumps or separated out?

fast_epochs <- EVA_Patients_5min %>% 
  filter(epoch_type == "fast epoch", !is.na(time_from_gastric_emptying_minute)) %>%
  arrange(combined_ID, time_from_gastric_emptying_minute) %>%
  group_by(combined_ID) %>%
  mutate(time_gap = time_from_gastric_emptying_minute - lag(time_from_gastric_emptying_minute) - 5) %>%
  ungroup()

fast_epochs <- fast_epochs %>%
  group_by(combined_ID) %>%
  summarize(
    n_fast_epochs = n(),
    avg_gap = round(mean(time_gap, na.rm = TRUE),2),
    median_gap = round(median(time_gap, na.rm = TRUE),2),
    min_gap = round(min(time_gap, na.rm = TRUE),2),
    max_gap = round(max(time_gap, na.rm = TRUE),2)
  )

medium_epochs <- EVA_Patients_5min %>% 
  filter(epoch_type == "medium epoch", !is.na(time_from_gastric_emptying_minute)) %>%
  arrange(combined_ID, time_from_gastric_emptying_minute) %>%
  group_by(combined_ID) %>%
  mutate(time_gap = time_from_gastric_emptying_minute - lag(time_from_gastric_emptying_minute)) %>%
  ungroup()

medium_epochs <- medium_epochs %>%
  group_by(combined_ID) %>%
  summarize(
    n_medium_epochs = n(),
    avg_gap = round(mean(time_gap, na.rm = TRUE),2),
    median_gap = round(median(time_gap, na.rm = TRUE),2),
    min_gap = round(min(time_gap, na.rm = TRUE),2),
    max_gap = round(max(time_gap, na.rm = TRUE),2)
  )

slow_epochs <- EVA_Patients_5min %>% 
  filter(epoch_type == "slow epoch", !is.na(time_from_gastric_emptying_minute)) %>%
  arrange(combined_ID, time_from_gastric_emptying_minute) %>%
  group_by(combined_ID) %>%
  mutate(time_gap = time_from_gastric_emptying_minute - lag(time_from_gastric_emptying_minute)) %>%
  ungroup()

slow_epochs <- slow_epochs %>%
  group_by(combined_ID) %>%
  summarize(
    n_slow_epochs = n(),
    avg_gap = round(mean(time_gap, na.rm = TRUE),2),
    median_gap = round(median(time_gap, na.rm = TRUE),2),
    min_gap = round(min(time_gap, na.rm = TRUE),2),
    max_gap = round(max(time_gap, na.rm = TRUE),2)
  )

2.3 Epoch Type - Time Fraction Plot

EVA_Patients_5min <- EVA_Patients_5min %>% 
  group_by(combined_ID) %>%
  mutate(time_fraction_end = time_fraction + 5/60/total_time)

# ggplot(EVA_Patients_5min) +
#   geom_segment(aes(x = time_fraction, xend = time_fraction_end, 
#                    y = as.character(combined_ID), yend = as.character(combined_ID), color = epoch_type)) +
#   labs(x = "Time Fraction", y = "ID", color = "Epoch Type") +
#   theme_minimal() +
#   theme(axis.text.y = element_text(size = 7))

EVA_Patients_5min$ID <- as.integer(factor(as.factor(EVA_Patients_5min$combined_ID)))
EVA_Patients_5min <- EVA_Patients_5min %>%
  mutate(ymin = ID - 0.5,
         ymax = ID + 0.5)

ggplot(EVA_Patients_5min  %>% mutate(
  epoch_type = case_when(
    epoch_type == "fast epoch" ~ "fast",
    epoch_type == "medium epoch" ~ "medium",
    epoch_type == "slow epoch" ~ "slow"
  )
)) +
  geom_rect(aes(xmin = time_fraction, xmax = time_fraction_end,
                ymin = ymin, ymax = ymax, fill = epoch_type)) +
  scale_y_continuous(
    breaks = EVA_Patients_5min$ID,
    labels = EVA_Patients_5min$combined_ID
  ) +
  labs(x = "Time Fraction since Gastric Emptying Time", y = "Participant ID", fill = "Type of Rhythmic Intervals") +
  theme_minimal() +
  theme(
    text = element_text(size = 14),
    axis.text.x = element_text(size = 14),
    axis.text.y = element_blank(),
    axis.title.x = element_text(size = 18),
    axis.title.y = element_text(size = 18),
    plot.title = element_text(size = 16)
  ) +   
  geom_segment(
    aes(x = -Inf, xend = Inf, y = ymin, yend = ymin),
    color = "black", linewidth = 0.02
  )

2.4 Is Total Time Correlated to Number of Epochs?

library(corrplot)
res <- cor(data.frame(
  total_time = (EVA_Patients_5min %>% select(combined_ID, total_time, total_no_missing_time, number_of_total_epochs, number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs) %>% unique())$total_time,
  total_no_missing_time = (EVA_Patients_5min %>% select(combined_ID, total_time, total_no_missing_time, number_of_total_epochs, number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs) %>% unique())$total_no_missing_time,
  number_of_total_epochs = (EVA_Patients_5min %>% select(combined_ID, total_time, total_no_missing_time, number_of_total_epochs, number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs) %>% unique())$number_of_total_epochs,
  number_of_fast_epochs = (EVA_Patients_5min %>% select(combined_ID, total_time, total_no_missing_time, number_of_total_epochs, number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs) %>% unique())$number_of_fast_epochs,
  number_of_medium_epochs = (EVA_Patients_5min %>% select(combined_ID, total_time, total_no_missing_time, number_of_total_epochs, number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs) %>% unique())$number_of_medium_epochs,
  number_of_slow_epochs = (EVA_Patients_5min %>% select(combined_ID, total_time, total_no_missing_time, number_of_total_epochs, number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs) %>% unique())$number_of_slow_epochs
  ), 
  use = "complete.obs", method = "spearman")
# res
corrplot(res, type = "upper", 
         tl.col = "black", tl.srt = 45)

3 EVA Controls Analyses

EVA_Controls_5min <- read.csv("C:/Users/zhaoz03/OneDrive - The Mount Sinai Hospital/MY PROJECTS/Project 11. GI Pressue EVA Study/Verified Dataset/EVA_Controls_5min.csv")


EVA_Controls_5min <- EVA_Controls_5min %>%
  filter(!is.na(epoch_type)) %>% 
  group_by(combined_ID) %>% 
  mutate(number_of_fast_epochs = sum(epoch_type == "fast epoch"), number_of_medium_epochs = sum(epoch_type == "medium epoch"), number_of_slow_epochs = sum(epoch_type == "slow epoch")) %>%
  ungroup() %>% 
  mutate(
    number_of_total_epochs = rowSums(across(c(number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs))),
    number_of_fast_epochs_frequency = number_of_fast_epochs/total_no_missing_time,
    number_of_medium_epochs_frequency = number_of_medium_epochs/total_no_missing_time,
    number_of_slow_epochs_frequency = number_of_slow_epochs/total_no_missing_time,
    number_of_total_epochs_frequency = number_of_total_epochs/total_no_missing_time)


EVA_Controls_5min <- EVA_Controls_5min %>% filter(full_window_size == "Yes", !is.na(epoch_type)) 
EVA_Controls_5min <- EVA_Controls_5min %>% filter(prefiltration == FALSE)

EVA_Controls_5min <- EVA_Controls_5min %>% 
  mutate(time_from_swallow_hour = time_from_swallow_minute/60,
         time_from_gastric_emptying_hour = time_from_gastric_emptying_minute/60)

EVA_Controls_5min <- EVA_Controls_5min %>%
  mutate(time_fraction = time_from_gastric_emptying_hour/total_time)

EVA_Controls_5min <- EVA_Controls_5min %>%
  rename(sum_of_amplitude = total_pressure_exposure,
        imputed_sum_of_amplitude = imputed_total_pressure_exposure)

EVA_Controls_5min <- EVA_Controls_5min %>%
  mutate(motility_index = log(sum_of_amplitude * total_number_of_contractions + 1))

3.1 Describe within-subject RMSSD variability for Control Data

Question: Does it seem true that for fast epochs RMSSD is more consistent?

Answer: It seems true that faster epochs RMSSD is more consistent, especially for RMSSD Time.

rmssd_variability_EVA_Controls_5min <- EVA_Controls_5min %>%
  group_by(combined_ID, epoch_type) %>%
  summarize(
    mean_RMSSD_Time = mean(epoch_RMSSD_Time),
    median_RMSSD_Time = median(epoch_RMSSD_Time),
    sd_RMSSD_Time = sd(epoch_RMSSD_Time),
    mean_RMSSD_Magnitude = mean(epoch_RMSSD_Magnitude),
    median_RMSSD_Magnitude = median(epoch_RMSSD_Magnitude),
    sd_RMSSD_Magnitude = sd(epoch_RMSSD_Magnitude),
    n_epochs = n(),
    .groups = "drop"
  )

p1 <- ggplot(rmssd_variability_EVA_Controls_5min, aes(x = epoch_type, y = sd_RMSSD_Time)) +
  geom_boxplot(outlier.shape = NA) +
  geom_jitter(width = 0.2, alpha = 0.5) +
  labs(
    title = "Standard Deviation of RMSSD Time for each ID by Epoch Type",
    x = "Epoch Type",
    y = "SD of RMSSD Time"
  ) +
  theme_minimal()

p2 <- ggplot(rmssd_variability_EVA_Controls_5min, aes(x = epoch_type, y = sd_RMSSD_Magnitude)) +
  geom_boxplot(outlier.shape = NA) +
  geom_jitter(width = 0.2, alpha = 0.5) +
  labs(
    title = "Standard Deviation of RMSSD Magnitude for each ID by Epoch Type",
    x = "Epoch Type",
    y = "SD of RMSSD Magnitude"
  ) +
  theme_minimal()

p3 <- ggplot(rmssd_variability_EVA_Controls_5min, aes(x = epoch_type, y = median_RMSSD_Time)) +
  geom_boxplot(outlier.shape = NA) +
  geom_jitter(width = 0.2, alpha = 0.5) +
  labs(
    title = "Median of RMSSD Time for each ID by Epoch Type",
    x = "Epoch Type",
    y = "Median of RMSSD Time"
  ) +
  theme_minimal()

p4 <- ggplot(rmssd_variability_EVA_Controls_5min, aes(x = epoch_type, y = median_RMSSD_Magnitude)) +
  geom_boxplot(outlier.shape = NA) +
  geom_jitter(width = 0.2, alpha = 0.5) +
  labs(
    title = "Median of RMSSD Magnitude for each ID by Epoch Type",
    x = "Epoch Type",
    y = "Median of RMSSD Magnitude"
  ) +
  theme_minimal()

grid.arrange(p1, p2, p3, p4, ncol = 2)

mycontrols <- tableby.control(test = TRUE, total = TRUE,
                              numeric.test = "kwt", cat.test = "chisq",
                              numeric.stats=c("N", "Nmiss", "meansd", "medianq1q3", "min", "max"),
                              cat.stats=c("countpct"),
                              stats.labels = list(
    meansd = "Mean (SD)",
    # medianq1q3 = "Median (Q1, Q3)",
    # range = "Min - Max",
    quantile5 = "5th Percentile",
    quantile95 = "95th Percentile",
    Nmiss2 = "Missing"
  ))

tab.test <- tableby(epoch_type ~ mean_RMSSD_Time + median_RMSSD_Time + sd_RMSSD_Time + mean_RMSSD_Magnitude + median_RMSSD_Magnitude + sd_RMSSD_Magnitude, data = rmssd_variability_EVA_Controls_5min, control=mycontrols) 

summary(tab.test)

	fast epoch (N=11)	medium epoch (N=8)	slow epoch (N=10)	Total (N=29)	p value
mean_RMSSD_Time					< 0.001
N	11	8	10	29
Mean (SD)	53.880 (5.451)	61.980 (7.329)	89.182 (19.362)	68.288 (19.834)
Median (Q1, Q3)	53.760 (49.975, 57.568)	63.686 (57.229, 66.366)	84.019 (80.527, 95.988)	63.740 (54.920, 80.514)
Min	46.780	49.140	68.541	46.780
Max	64.695	72.147	135.417	135.417
median_RMSSD_Time					< 0.001
N	11	8	10	29
Mean (SD)	52.868 (6.292)	61.939 (7.267)	83.365 (15.330)	65.887 (16.824)
Median (Q1, Q3)	53.760 (48.420, 56.660)	63.678 (57.229, 66.366)	80.938 (73.425, 86.995)	63.740 (54.920, 71.835)
Min	42.750	49.140	65.970	42.750
Max	64.695	71.835	116.840	116.840
sd_RMSSD_Time					0.123
N	7	6	9	22
N-Miss	4	2	1	7
Mean (SD)	11.714 (5.622)	8.811 (4.061)	26.624 (20.648)	17.022 (15.606)
Median (Q1, Q3)	12.554 (7.798, 16.008)	8.637 (6.642, 10.105)	20.541 (8.871, 38.521)	11.506 (7.605, 17.869)
Min	3.665	3.550	6.433	3.550
Max	18.166	15.473	65.530	65.530
mean_RMSSD_Magnitude					0.041
N	11	8	10	29
Mean (SD)	96.743 (51.984)	55.823 (20.957)	51.975 (23.124)	70.017 (41.246)
Median (Q1, Q3)	83.154 (50.300, 141.909)	53.901 (38.620, 71.942)	45.989 (35.049, 58.565)	53.913 (41.480, 83.154)
Min	42.564	28.360	27.810	27.810
Max	189.670	83.405	101.350	189.670
median_RMSSD_Magnitude					0.071
N	11	8	10	29
Mean (SD)	84.860 (48.989)	55.849 (21.074)	49.098 (24.307)	64.525 (37.776)
Median (Q1, Q3)	69.730 (50.300, 107.558)	53.620 (38.620, 72.134)	41.740 (35.720, 53.392)	52.460 (38.900, 72.360)
Min	36.570	28.360	28.680	28.360
Max	189.670	83.405	112.800	189.670
sd_RMSSD_Magnitude					0.461
N	7	6	9	22
N-Miss	4	2	1	7
Mean (SD)	49.549 (52.164)	20.166 (14.553)	20.638 (17.663)	29.709 (33.752)
Median (Q1, Q3)	37.371 (12.714, 60.335)	18.262 (8.720, 27.837)	16.352 (11.450, 19.040)	17.468 (10.933, 35.443)
Min	8.980	4.936	7.005	4.936
Max	154.397	42.971	65.932	154.397

knitr::kable(rmssd_variability_EVA_Controls_5min %>%
               group_by(epoch_type) %>%
                summarize(
                  p5_sd_RMSSD_Time = quantile(sd_RMSSD_Time, 0.05, na.rm = TRUE),
                  p5_median_RMSSD_Time = quantile(median_RMSSD_Time, 0.05, na.rm = TRUE),
                  p5_mean_RMSSD_Time = quantile(mean_RMSSD_Time, 0.05, na.rm = TRUE),
                  p95_sd_RMSSD_Time = quantile(sd_RMSSD_Time, 0.95, na.rm = TRUE),
                  p95_median_RMSSD_Time = quantile(median_RMSSD_Time, 0.95, na.rm = TRUE),
                  p95_mean_RMSSD_Time = quantile(mean_RMSSD_Time, 0.95, na.rm = TRUE),
                  p5_sd_RMSSD_Magnitude = quantile(sd_RMSSD_Magnitude, 0.05, na.rm = TRUE),
                  p5_median_RMSSD_Magnitude = quantile(median_RMSSD_Magnitude, 0.05, na.rm = TRUE),
                  p5_mean_RMSSD_Magnitude = quantile(mean_RMSSD_Magnitude, 0.05, na.rm = TRUE),
                  p95_sd_RMSSD_Magnitude = quantile(sd_RMSSD_Magnitude, 0.95, na.rm = TRUE),
                  p95_median_RMSSD_Magnitude = quantile(median_RMSSD_Magnitude, 0.95, na.rm = TRUE),
                  p95_mean_RMSSD_Magnitude = quantile(mean_RMSSD_Magnitude, 0.95, na.rm = TRUE),
                ), 
  format = "html")

epoch_type	p5_sd_RMSSD_Time	p5_median_RMSSD_Time	p5_mean_RMSSD_Time	p95_sd_RMSSD_Time	p95_median_RMSSD_Time	p95_mean_RMSSD_Time	p5_sd_RMSSD_Magnitude	p5_median_RMSSD_Magnitude	p5_mean_RMSSD_Magnitude	p95_sd_RMSSD_Magnitude	p95_median_RMSSD_Magnitude	p95_mean_RMSSD_Magnitude
fast epoch	4.137823	44.62000	47.08500	17.80919	61.5675	61.57000	9.257187	41.2700	43.69814	131.27503	169.38000	172.08300
medium epoch	4.190466	51.53925	51.53925	14.21916	70.3475	70.55008	5.428812	31.6570	31.65700	39.64327	83.04975	82.78083
slow epoch	6.818168	66.56400	69.55779	58.76796	108.7580	120.26817	8.507823	29.6565	30.95100	48.68152	87.29400	91.46050

p1 <- ggplot(EVA_Controls_5min %>% mutate(new_ID = paste0(combined_ID, epoch_type)) %>% filter(epoch_type == "fast epoch"), aes(x = factor(new_ID), y = epoch_RMSSD_Time)) +
  geom_boxplot() +
  labs(
    title = "RMSSD Time Distribution by Fast Epoch",
    x = "ID",
    y = "RMSSD Time"
  ) +
  ylim(0, 200) +
  theme_minimal()

p2 <- ggplot(EVA_Controls_5min %>% mutate(new_ID = paste0(combined_ID, epoch_type)) %>% filter(epoch_type == "medium epoch"), aes(x = factor(new_ID), y = epoch_RMSSD_Time)) +
  geom_boxplot() +
  labs(
    title = "RMSSD Time Distribution by Medium Epoch",
    x = "ID",
    y = "RMSSD Time"
  ) +
  ylim(0, 200) +
  theme_minimal()

p3 <- ggplot(EVA_Controls_5min %>% mutate(new_ID = paste0(combined_ID, epoch_type)) %>% filter(epoch_type == "slow epoch"), aes(x = factor(new_ID), y = epoch_RMSSD_Time)) +
  geom_boxplot() +
  labs(
    title = "RMSSD Time Distribution by Slow Epoch",
    x = "ID",
    y = "RMSSD Time"
  ) +
  ylim(0, 200) +
  theme_minimal()

grid.arrange(p1, p2, p3, ncol = 3)

p1 <- ggplot(EVA_Controls_5min %>% mutate(new_ID = paste0(combined_ID, epoch_type)) %>% filter(epoch_type == "fast epoch"), aes(x = factor(new_ID), y = epoch_RMSSD_Magnitude)) +
  geom_boxplot() +
  labs(
    title = "RMSSD Magnitude Distribution by Fast Epoch",
    x = "ID",
    y = "RMSSD Magnitude"
  ) +
  ylim(0, 400) +
  theme_minimal()

p2 <- ggplot(EVA_Controls_5min %>% mutate(new_ID = paste0(combined_ID, epoch_type)) %>% filter(epoch_type == "medium epoch"), aes(x = factor(new_ID), y = epoch_RMSSD_Magnitude)) +
  geom_boxplot() +
  labs(
    title = "RMSSD Magnitude Distribution by Medium Epoch",
    x = "ID",
    y = "RMSSD Magnitude"
  ) +
  ylim(0, 400) +
  theme_minimal()

p3 <- ggplot(EVA_Controls_5min %>% mutate(new_ID = paste0(combined_ID, epoch_type)) %>% filter(epoch_type == "slow epoch"), aes(x = factor(new_ID), y = epoch_RMSSD_Magnitude)) +
  geom_boxplot() +
  labs(
    title = "RMSSD Magnitude Distribution by Slow Epoch",
    x = "ID",
    y = "RMSSD Magnitude"
  ) +
  ylim(0, 400) +
  theme_minimal()

grid.arrange(p1, p2, p3, ncol = 3)

3.2 Describe how the epochs are distributed over time

Question: Are they more in the proximal or distal small bowel?

p1 <- ggplot(EVA_Controls_5min %>% filter(epoch_type == "fast epoch"), aes(x = time_fraction)) +
  geom_histogram(fill = "steelblue", color = "white") +
  labs(
    title = "Distribution of Fast Epochs Over Time Fraction Since Gastric Emptying",
    x = "Time Fraction",
    y = "Number of Epochs"
  ) +
  theme_minimal()

p2 <- ggplot(EVA_Controls_5min %>% filter(epoch_type == "medium epoch"), aes(x = time_fraction)) +
  geom_histogram(fill = "steelblue", color = "white") +
  labs(
    title = "Distribution of Medium Epochs Over Time Fraction Since Gastric Emptying",
    x = "Time Fraction",
    y = "Number of Epochs"
  ) +
  theme_minimal()

p3 <- ggplot(EVA_Controls_5min %>% filter(epoch_type == "slow epoch"), aes(x = time_fraction)) +
  geom_histogram(fill = "steelblue", color = "white") +
  labs(
    title = "Distribution of Slow Epochs Over Time Fraction Since Gastric Emptying",
    x = "Time Fraction",
    y = "Number of Epochs"
  ) +
  theme_minimal()

grid.arrange(p1, p2, p3, ncol = 3)

Question: Do they tend to occur together in clumps or separated out?

fast_epochs <- EVA_Controls_5min %>% 
  filter(epoch_type == "fast epoch", !is.na(time_from_gastric_emptying_minute)) %>%
  arrange(combined_ID, time_from_gastric_emptying_minute) %>%
  group_by(combined_ID) %>%
  mutate(time_gap = time_from_gastric_emptying_minute - lag(time_from_gastric_emptying_minute) - 5) %>%
  ungroup()

fast_epochs <- fast_epochs %>%
  group_by(combined_ID) %>%
  summarize(
    n_fast_epochs = n(),
    avg_gap = round(mean(time_gap, na.rm = TRUE),2),
    median_gap = round(median(time_gap, na.rm = TRUE),2),
    min_gap = round(min(time_gap, na.rm = TRUE),2),
    max_gap = round(max(time_gap, na.rm = TRUE),2)
  )

medium_epochs <- EVA_Controls_5min %>% 
  filter(epoch_type == "medium epoch", !is.na(time_from_gastric_emptying_minute)) %>%
  arrange(combined_ID, time_from_gastric_emptying_minute) %>%
  group_by(combined_ID) %>%
  mutate(time_gap = time_from_gastric_emptying_minute - lag(time_from_gastric_emptying_minute)) %>%
  ungroup()

medium_epochs <- medium_epochs %>%
  group_by(combined_ID) %>%
  summarize(
    n_medium_epochs = n(),
    avg_gap = round(mean(time_gap, na.rm = TRUE),2),
    median_gap = round(median(time_gap, na.rm = TRUE),2),
    min_gap = round(min(time_gap, na.rm = TRUE),2),
    max_gap = round(max(time_gap, na.rm = TRUE),2)
  )

slow_epochs <- EVA_Controls_5min %>% 
  filter(epoch_type == "slow epoch", !is.na(time_from_gastric_emptying_minute)) %>%
  arrange(combined_ID, time_from_gastric_emptying_minute) %>%
  group_by(combined_ID) %>%
  mutate(time_gap = time_from_gastric_emptying_minute - lag(time_from_gastric_emptying_minute)) %>%
  ungroup()

slow_epochs <- slow_epochs %>%
  group_by(combined_ID) %>%
  summarize(
    n_slow_epochs = n(),
    avg_gap = round(mean(time_gap, na.rm = TRUE),2),
    median_gap = round(median(time_gap, na.rm = TRUE),2),
    min_gap = round(min(time_gap, na.rm = TRUE),2),
    max_gap = round(max(time_gap, na.rm = TRUE),2)
  )

3.3 Epoch Type - Time Fraction Plot

EVA_Controls_5min <- EVA_Controls_5min %>% 
  group_by(combined_ID) %>%
  mutate(time_fraction_end = time_fraction + 5/60/total_time)

# ggplot(EVA_Controls_5min) +
#   geom_segment(aes(x = time_fraction, xend = time_fraction_end, 
#                    y = as.character(combined_ID), yend = as.character(combined_ID), color = epoch_type)) +
#   labs(x = "Time Fraction", y = "ID", color = "Epoch Type") +
#   theme_minimal() +
#   theme(axis.text.y = element_text(size = 7))

EVA_Controls_5min$ID <- as.integer(factor(as.factor(EVA_Controls_5min$combined_ID)))
EVA_Controls_5min <- EVA_Controls_5min %>%
  mutate(ymin = ID - 0.5,
         ymax = ID + 0.5)

ggplot(EVA_Controls_5min %>% mutate(
  epoch_type = case_when(
    epoch_type == "fast epoch" ~ "fast",
    epoch_type == "medium epoch" ~ "medium",
    epoch_type == "slow epoch" ~ "slow"
  )
)) +
  geom_rect(aes(xmin = time_fraction, xmax = time_fraction_end,
                ymin = ymin, ymax = ymax, fill = epoch_type)) +
  scale_y_continuous(
    breaks = EVA_Controls_5min$ID,
    labels = EVA_Controls_5min$combined_ID
  ) +
  labs(x = "Time Fraction since Gastric Emptying Time", y = "Participant ID", fill = "Type of Rhythmic Intervals") +
  theme(
    text = element_text(size = 14),
    axis.text.x = element_text(size = 14),
    axis.text.y = element_blank(),
    axis.title.x = element_text(size = 18),
    axis.title.y = element_text(size = 18),
    plot.title = element_text(size = 16)
  ) +   
  geom_segment(
    aes(x = -Inf, xend = Inf, y = ymin, yend = ymin),
    color = "black", linewidth = 0.02
  )

3.4 Is Total Time Correlated to Number of Epochs?

library(corrplot)
res <- cor(data.frame(
  total_time = (EVA_Controls_5min %>% select(combined_ID, total_time, total_no_missing_time, number_of_total_epochs, number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs) %>% unique())$total_time,
  total_no_missing_time = (EVA_Controls_5min %>% select(combined_ID, total_time, total_no_missing_time, number_of_total_epochs, number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs) %>% unique())$total_no_missing_time,
  number_of_total_epochs = (EVA_Controls_5min %>% select(combined_ID, total_time, total_no_missing_time, number_of_total_epochs, number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs) %>% unique())$number_of_total_epochs,
  number_of_fast_epochs = (EVA_Controls_5min %>% select(combined_ID, total_time, total_no_missing_time, number_of_total_epochs, number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs) %>% unique())$number_of_fast_epochs,
  number_of_medium_epochs = (EVA_Controls_5min %>% select(combined_ID, total_time, total_no_missing_time, number_of_total_epochs, number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs) %>% unique())$number_of_medium_epochs,
  number_of_slow_epochs = (EVA_Controls_5min %>% select(combined_ID, total_time, total_no_missing_time, number_of_total_epochs, number_of_fast_epochs, number_of_medium_epochs, number_of_slow_epochs) %>% unique())$number_of_slow_epochs
  ), 
  use = "complete.obs", method = "spearman")
# res
corrplot(res, type = "upper", 
         tl.col = "black", tl.srt = 45)

# write.csv(fast_epochs, "fast_epochs.csv")
# write.csv(medium_epochs, "medium_epochs.csv")
# write.csv(slow_epochs, "slow_epochs.csv")

# EVA_Patients_5min <- read.csv("C:/Users/zhaoz03/OneDrive - The Mount Sinai Hospital/MY PROJECTS/Project 11. GI Pressue EVA Study/Dataset/EVA_Patients_5min.csv")
# EVA_Patients_5min <- EVA_Patients_5min %>% filter(full_window_size == "Yes", !is.na(epoch_type))

# library(readxl)
# meals <- read_excel("C:/Users/zhaoz03/OneDrive - The Mount Sinai Hospital/MY PROJECTS/Project 11. GI Pressue EVA Study/PatientEvents_summary_04292025.xlsx", sheet = 1)
# meals <- meals %>% filter(event_name == "Ate meal") %>% select(-c(note, event_name))
# meals$patient_id <- as.integer(meals$patient_id)

4 Tables

4.1 Table 1

Control_Patients_5min$source <- "Jack's Data"
EVA_Patients_5min$source <- "EVA Patient"
# EVA_Controls_5min$source <- "EVA Controls"

all_data <- rbind(Control_Patients_5min, EVA_Patients_5min)
all_data <- all_data %>% group_by(combined_ID) %>% slice(1)

Plot_conditioning(all_data, c("total_time", "total_no_missing_time", "no_missing_time_ratio", "total_number_of_contractions", "total_contractions_frequency", "sum_of_amplitude", "total_number_of_contractions_over26.7", "total_contraction_rate_0", "total_contraction_rate_1_3", "total_contraction_rate_4_6", "total_contraction_rate_7_9", "total_contraction_rate_over10", "number_of_contractions_over26.7_per_hour", "imputed_total_number_of_contractions", "imputed_sum_of_amplitude", "number_of_fast_epochs", "number_of_medium_epochs", "number_of_slow_epochs", "number_of_total_epochs", "number_of_fast_epochs_frequency", "number_of_medium_epochs_frequency", "number_of_slow_epochs_frequency", "number_of_total_epochs_frequency", "motility_index"), "source")

mycontrols <- tableby.control(test = T, total = TRUE,
                              numeric.test = "kwt", cat.test = "chisq",
                              numeric.stats=c("N", "Nmiss", "mean", "sd", "median", "q1q3", "min", "max"),
                              cat.stats=c("countpct"),
                              stats.labels=list(N='Count', Nmiss='Missing Values', mean='mean', sd='SD', median='Median', q1q3='Q1,Q3', min='Min', max='Max'))


tab.test <- tableby(source ~ total_time + total_no_missing_time + no_missing_time_ratio + total_number_of_contractions + total_contractions_frequency + sum_of_amplitude + total_number_of_contractions_over26.7 + total_contraction_rate_0 + total_contraction_rate_1_3 + total_contraction_rate_4_6 + total_contraction_rate_7_9 + total_contraction_rate_over10 + number_of_contractions_over26.7_per_hour + imputed_total_number_of_contractions + imputed_sum_of_amplitude + number_of_fast_epochs + number_of_medium_epochs + number_of_slow_epochs + number_of_total_epochs + number_of_fast_epochs_frequency + number_of_medium_epochs_frequency + number_of_slow_epochs_frequency + number_of_total_epochs_frequency + motility_index, data=all_data, control = mycontrols)

summary(tab.test)

	EVA Patient (N=71)	Jack’s Data (N=98)	Total (N=169)	p value
total_time				< 0.001
Count	71	98	169
mean	5.191	3.762	4.363
SD	3.099	1.160	2.297
Median	4.703	3.706	3.951
Q1,Q3	3.716, 5.750	2.910, 4.409	3.245, 4.973
Min	1.664	1.312	1.312
Max	23.128	6.590	23.128
total_no_missing_time				0.226
Count	71	98	169
mean	3.393	3.518	3.466
SD	1.579	1.044	1.293
Median	3.063	3.428	3.338
Q1,Q3	2.425, 4.372	2.870, 4.104	2.603, 4.170
Min	1.010	1.307	1.010
Max	8.811	6.212	8.811
no_missing_time_ratio				< 0.001
Count	71	98	169
mean	0.716	0.942	0.847
SD	0.241	0.084	0.202
Median	0.766	0.971	0.941
Q1,Q3	0.613, 0.897	0.939, 0.990	0.788, 0.981
Min	0.103	0.442	0.103
Max	0.997	1.000	1.000
total_number_of_contractions				0.012
Count	71	98	169
mean	973.676	786.857	865.343
SD	477.890	392.159	438.755
Median	885.000	734.000	780.000
Q1,Q3	615.000, 1239.500	517.500, 951.750	554.000, 1073.000
Min	252.000	174.000	174.000
Max	2582.000	2517.000	2582.000
total_contractions_frequency				< 0.001
Count	71	98	169
mean	312.995	226.842	263.037
SD	150.872	91.448	127.011
Median	285.281	218.881	254.833
Q1,Q3	224.779, 367.898	160.092, 272.262	174.544, 328.726
Min	118.919	52.572	52.572
Max	1122.063	433.827	1122.063
sum_of_amplitude				0.051
Count	71	98	169
mean	115979.438	100940.428	107258.592
SD	49744.617	37378.788	43510.713
Median	106323.294	96916.642	101003.974
Q1,Q3	82820.367, 148375.073	76487.509, 125008.950	77963.451, 132296.825
Min	34593.705	31958.930	31958.930
Max	262383.401	224859.417	262383.401
total_number_of_contractions_over26.7				< 0.001
Count	71	98	169
mean	136.352	90.847	109.964
SD	94.432	68.169	83.103
Median	115.000	72.000	89.000
Q1,Q3	59.000, 187.500	41.000, 127.500	47.000, 151.000
Min	8.000	12.000	8.000
Max	441.000	338.000	441.000
total_contraction_rate_0				0.493
Count	71	98	169
mean	0.237	0.247	0.243
SD	0.168	0.153	0.159
Median	0.216	0.222	0.218
Q1,Q3	0.129, 0.293	0.122, 0.341	0.129, 0.326
Min	0.012	0.000	0.000
Max	0.970	0.685	0.970
total_contraction_rate_1_3				0.274
Count	71	98	169
mean	0.327	0.344	0.337
SD	0.087	0.093	0.090
Median	0.336	0.357	0.343
Q1,Q3	0.268, 0.389	0.272, 0.411	0.272, 0.401
Min	0.029	0.110	0.029
Max	0.488	0.566	0.566
total_contraction_rate_4_6				0.866
Count	71	98	169
mean	0.188	0.187	0.188
SD	0.065	0.071	0.068
Median	0.189	0.192	0.189
Q1,Q3	0.137, 0.227	0.133, 0.225	0.133, 0.226
Min	0.000	0.041	0.000
Max	0.366	0.364	0.366
total_contraction_rate_7_9				0.836
Count	71	98	169
mean	0.130	0.136	0.133
SD	0.071	0.082	0.077
Median	0.112	0.115	0.114
Q1,Q3	0.085, 0.165	0.081, 0.192	0.082, 0.186
Min	0.000	0.005	0.000
Max	0.336	0.455	0.455
total_contraction_rate_over10				0.010
Count	71	98	169
mean	0.119	0.085	0.099
SD	0.091	0.071	0.081
Median	0.106	0.067	0.078
Q1,Q3	0.048, 0.160	0.030, 0.125	0.038, 0.143
Min	0.000	0.000	0.000
Max	0.405	0.283	0.405
number_of_contractions_over26.7_per_hour				< 0.001
Count	71	98	169
mean	46.195	27.688	35.463
SD	35.738	23.529	30.590
Median	38.169	20.876	25.165
Q1,Q3	20.018, 59.233	11.757, 37.625	15.186, 47.064
Min	4.460	3.119	3.119
Max	168.976	155.521	168.976
imputed_total_number_of_contractions				< 0.001
Count	71	98	169
mean	1561.901	842.918	1144.976
SD	1075.089	398.189	836.553
Median	1233.000	788.000	963.000
Q1,Q3	955.000, 1826.000	577.000, 1006.000	673.000, 1336.000
Min	332.000	204.000	204.000
Max	6210.000	2526.000	6210.000
imputed_sum_of_amplitude				< 0.001
Count	71	98	169
mean	187578.734	108176.988	141535.118
SD	139333.716	38328.935	102383.712
Median	158548.129	103445.589	119039.990
Q1,Q3	115931.473, 200990.440	79719.797, 129270.590	89727.664, 160351.580
Min	62203.157	34829.877	34829.877
Max	950697.962	225832.269	950697.962
number_of_fast_epochs				0.361
Count	71	98	169
mean	4.113	4.724	4.467
SD	3.871	4.497	4.244
Median	3.000	4.000	4.000
Q1,Q3	1.000, 6.000	1.000, 7.000	1.000, 7.000
Min	0.000	0.000	0.000
Max	17.000	27.000	27.000
number_of_medium_epochs				0.013
Count	71	98	169
mean	2.592	3.531	3.136
SD	2.950	2.894	2.946
Median	2.000	3.000	2.000
Q1,Q3	1.000, 3.000	1.000, 5.000	1.000, 5.000
Min	0.000	0.000	0.000
Max	18.000	13.000	18.000
number_of_slow_epochs				0.006
Count	71	98	169
mean	3.437	4.663	4.148
SD	3.379	3.386	3.427
Median	3.000	4.000	3.000
Q1,Q3	1.000, 5.000	2.000, 6.000	1.000, 6.000
Min	0.000	0.000	0.000
Max	15.000	15.000	15.000
number_of_total_epochs				0.011
Count	71	98	169
mean	10.141	12.918	11.751
SD	7.769	8.196	8.113
Median	9.000	12.000	10.000
Q1,Q3	5.000, 14.000	7.000, 18.000	6.000, 17.000
Min	1.000	1.000	1.000
Max	37.000	51.000	51.000
number_of_fast_epochs_frequency				0.651
Count	71	98	169
mean	1.275	1.377	1.334
SD	1.138	1.188	1.165
Median	1.115	1.108	1.115
Q1,Q3	0.395, 1.765	0.378, 2.274	0.384, 1.931
Min	0.000	0.000	0.000
Max	5.168	4.601	5.168
number_of_medium_epochs_frequency				0.027
Count	71	98	169
mean	0.729	1.019	0.897
SD	0.654	0.837	0.777
Median	0.622	0.847	0.743
Q1,Q3	0.300, 0.994	0.339, 1.529	0.315, 1.268
Min	0.000	0.000	0.000
Max	3.240	4.302	4.302
number_of_slow_epochs_frequency				0.002
Count	71	98	169
mean	0.899	1.295	1.129
SD	0.722	0.850	0.820
Median	0.779	1.214	0.980
Q1,Q3	0.366, 1.406	0.689, 1.832	0.497, 1.649
Min	0.000	0.000	0.000
Max	3.232	3.648	3.648
number_of_total_epochs_frequency				0.008
Count	71	98	169
mean	2.903	3.691	3.360
SD	1.803	2.027	1.970
Median	2.512	3.430	3.091
Q1,Q3	1.689, 3.855	2.159, 4.985	1.779, 4.427
Min	0.326	0.302	0.302
Max	7.846	9.386	9.386
motility_index				0.023
Count	71	98	169
mean	18.329	18.001	18.139
SD	0.928	0.873	0.908
Median	18.397	18.086	18.147
Q1,Q3	17.833, 19.057	17.521, 18.615	17.589, 18.737
Min	15.981	15.750	15.750
Max	20.278	20.154	20.278

Control_Patients_5min_RMSSD <- Control_Patients_5min %>% 
  group_by(combined_ID, epoch_type) %>%
  summarize(
    mean_RMSSD_Time = mean(epoch_RMSSD_Time),
    median_RMSSD_Time = median(epoch_RMSSD_Time),
    sd_RMSSD_Time = sd(epoch_RMSSD_Time),
    mean_RMSSD_Magnitude = mean(epoch_RMSSD_Magnitude),
    median_RMSSD_Magnitude = median(epoch_RMSSD_Magnitude),
    sd_RMSSD_Magnitude = sd(epoch_RMSSD_Magnitude),
    n_epochs = n(),
    .groups = "drop"
  ) %>%
  pivot_wider(
    id_cols = combined_ID,
    names_from = epoch_type,
    values_from = c(mean_RMSSD_Time, median_RMSSD_Time, sd_RMSSD_Time, mean_RMSSD_Magnitude, median_RMSSD_Magnitude, sd_RMSSD_Magnitude, n_epochs)) %>%
  mutate(source = "Jack's Data")

Control_Patients_5min_total_epoch <- Control_Patients_5min %>% 
  group_by(combined_ID) %>%
  summarize(
    median_RMSSD_Time_total_epoch = median(epoch_RMSSD_Time),
    median_RMSSD_Magnitude_total_epoch = median(epoch_RMSSD_Magnitude),
    .groups = "drop"
  )

Control_Patients_5min_RMSSD <- Control_Patients_5min_RMSSD %>%
  left_join(Control_Patients_5min_total_epoch)
    
EVA_Patients_5min_RMSSD <- EVA_Patients_5min %>% 
  group_by(combined_ID, epoch_type) %>%
  summarize(
    mean_RMSSD_Time = mean(epoch_RMSSD_Time),
    median_RMSSD_Time = median(epoch_RMSSD_Time),
    sd_RMSSD_Time = sd(epoch_RMSSD_Time),
    mean_RMSSD_Magnitude = mean(epoch_RMSSD_Magnitude),
    median_RMSSD_Magnitude = median(epoch_RMSSD_Magnitude),
    sd_RMSSD_Magnitude = sd(epoch_RMSSD_Magnitude),
    n_epochs = n(),
    .groups = "drop"
  ) %>%
  pivot_wider(
    id_cols = combined_ID,
    names_from = epoch_type,
    values_from = c(mean_RMSSD_Time, median_RMSSD_Time, sd_RMSSD_Time, mean_RMSSD_Magnitude, median_RMSSD_Magnitude, sd_RMSSD_Magnitude, n_epochs)
  )%>%
  mutate(source = "EVA Data")

EVA_Patients_5min_total_epoch <- EVA_Patients_5min %>% 
  group_by(combined_ID) %>%
  summarize(
    median_RMSSD_Time_total_epoch = median(epoch_RMSSD_Time),
    median_RMSSD_Magnitude_total_epoch = median(epoch_RMSSD_Magnitude),
    .groups = "drop"
  )

EVA_Patients_5min_RMSSD <- EVA_Patients_5min_RMSSD %>%
  left_join(EVA_Patients_5min_total_epoch)

all_data_RMSSD <- rbind(Control_Patients_5min_RMSSD, EVA_Patients_5min_RMSSD)

mycontrols <- tableby.control(test = T, total = TRUE,
                              numeric.test = "kwt", cat.test = "chisq",
                              numeric.stats=c("N", "Nmiss", "mean", "sd", "median", "q1q3", "min", "max"),
                              cat.stats=c("countpct"),
                              stats.labels=list(N='Count', Nmiss='Missing Values', mean='mean', sd='SD', median='Median', q1q3='Q1,Q3', min='Min', max='Max'))


tab.test <- tableby(source ~ `median_RMSSD_Time_fast epoch` + `median_RMSSD_Time_medium epoch` + `median_RMSSD_Time_slow epoch`+ median_RMSSD_Time_total_epoch + `median_RMSSD_Magnitude_fast epoch` + `median_RMSSD_Magnitude_medium epoch` + `median_RMSSD_Magnitude_slow epoch` + median_RMSSD_Magnitude_total_epoch + `n_epochs_fast epoch` + `n_epochs_medium epoch` + `n_epochs_slow epoch`, data=all_data_RMSSD, control = mycontrols)

summary(tab.test)

	EVA Data (N=71)	Jack’s Data (N=98)	Total (N=169)	p value
median_RMSSD_Time_fast epoch				0.012
Count	61	83	144
Missing Values	10	15	25
mean	53.136	50.465	51.597
SD	6.481	4.285	5.469
Median	52.540	49.280	50.237
Q1,Q3	48.160, 57.180	47.425, 53.165	47.701, 55.321
Min	41.540	43.860	41.540
Max	71.790	63.850	71.790
median_RMSSD_Time_medium epoch				0.005
Count	58	87	145
Missing Values	13	11	24
mean	67.132	63.418	64.904
SD	9.843	7.848	8.859
Median	65.590	61.950	63.230
Q1,Q3	60.888, 70.885	58.100, 65.892	59.030, 68.370
Min	50.480	52.110	50.480
Max	101.345	96.330	101.345
median_RMSSD_Time_slow epoch				0.042
Count	59	90	149
Missing Values	12	8	20
mean	83.876	78.458	80.604
SD	14.217	9.098	11.664
Median	80.715	77.975	79.480
Q1,Q3	73.755, 89.062	71.345, 84.540	72.840, 86.380
Min	64.010	52.200	52.200
Max	138.230	101.360	138.230
median_RMSSD_Time_total_epoch				0.968
Count	71	98	169
mean	64.233	64.060	64.133
SD	12.860	10.438	11.481
Median	63.615	62.395	63.110
Q1,Q3	57.370, 69.165	57.683, 70.236	57.460, 69.530
Min	46.380	46.620	46.380
Max	138.230	98.730	138.230
median_RMSSD_Magnitude_fast epoch				0.068
Count	61	83	144
Missing Values	10	15	25
mean	73.035	65.556	68.724
SD	28.061	24.956	26.482
Median	72.370	61.095	66.670
Q1,Q3	54.360, 86.700	47.792, 75.675	49.625, 82.172
Min	24.450	20.160	20.160
Max	158.900	157.550	158.900
median_RMSSD_Magnitude_medium epoch				0.249
Count	58	87	145
Missing Values	13	11	24
mean	60.919	56.492	58.263
SD	25.692	23.552	24.440
Median	53.445	49.800	51.210
Q1,Q3	42.785, 71.294	40.300, 63.960	40.840, 67.570
Min	27.860	28.345	27.860
Max	151.905	149.150	151.905
median_RMSSD_Magnitude_slow epoch				0.045
Count	59	90	149
Missing Values	12	8	20
mean	47.339	43.154	44.811
SD	24.965	26.142	25.679
Median	43.950	35.885	40.850
Q1,Q3	34.177, 49.892	26.372, 49.825	28.360, 50.255
Min	17.460	20.100	17.460
Max	177.435	180.510	180.510
median_RMSSD_Magnitude_total_epoch				0.010
Count	71	98	169
mean	59.947	52.018	55.349
SD	24.122	21.697	23.014
Median	55.440	47.945	51.170
Q1,Q3	45.390, 71.448	37.714, 59.303	39.895, 62.280
Min	22.125	21.700	21.700
Max	139.395	136.030	139.395
n_epochs_fast epoch				0.182
Count	61	83	144
Missing Values	10	15	25
mean	4.787	5.578	5.243
SD	3.769	4.370	4.131
Median	4.000	5.000	5.000
Q1,Q3	2.000, 7.000	2.500, 7.000	2.000, 7.000
Min	1.000	1.000	1.000
Max	17.000	27.000	27.000
n_epochs_medium epoch				0.031
Count	58	87	145
Missing Values	13	11	24
mean	3.172	3.977	3.655
SD	2.968	2.766	2.866
Median	2.500	3.000	3.000
Q1,Q3	1.000, 4.000	2.000, 6.000	2.000, 5.000
Min	1.000	1.000	1.000
Max	18.000	13.000	18.000
n_epochs_slow epoch				0.028
Count	59	90	149
Missing Values	12	8	20
mean	4.136	5.078	4.705
SD	3.293	3.219	3.270
Median	3.000	4.000	4.000
Q1,Q3	1.500, 5.500	3.000, 7.000	2.000, 6.000
Min	1.000	1.000	1.000
Max	15.000	15.000	15.000

4.2 Table 2

4.2.1 Compare GI syptoms between 3 EVA Groups

final_spreadsheet <- read_excel("C:/Users/zhaoz03/OneDrive - The Mount Sinai Hospital/MY PROJECTS/Project 11. GI Pressue EVA Study/Dataset/Final spreadsheet_controls and PWH.xlsx", sheet = 2, skip = 1)

# final_spreadsheet <- final_spreadsheet %>% filter(!is.na(SBTT...34), !is.na(cass_total))

final_spreadsheet <- final_spreadsheet %>%
  mutate(
    groups = case_when(
      as.numeric(SBTT...34) > 360 ~ "prolonged_SBTT",
      cass_total >= 3 & as.numeric(SBTT...34) <= 360 ~ "normal_SBTT_abnormal_CASS",
      cass_total < 3 & as.numeric(SBTT...34) <= 360 ~ "normal_SBTT_normal_CASS"
    ))

EVA_Patients_5min <- EVA_Patients_5min %>%
  left_join(final_spreadsheet %>% select(ID, ortho_intolerance_score, vasomotor_score, secretomotor_score, gi_score, bladder_score, pupillometer_score, total_compass_31, a_heartburn_regurgitation, b_fullness_early_satiety, c_nausea_vomiting, d_bloating, e_upper_abdominal_pain, f_lower_abdominal_pain, total_pagi_sym_score, groups), by = c("combined_ID" = "ID"))
Control_Patients_5min$groups <- "Jack's Data"
all_data <- rbind(Control_Patients_5min, EVA_Patients_5min)
all_data <- all_data %>% group_by(combined_ID) %>% slice(1)

mycontrols <- tableby.control(test = T, total = TRUE,
                              numeric.test = "kwt", cat.test = "chisq",
                              numeric.stats=c("N", "Nmiss", "mean", "sd", "median", "q1q3", "min", "max"),
                              cat.stats=c("countpct"),
                              stats.labels=list(N='Count', Nmiss='Missing Values', mean='mean', sd='SD', median='Median', q1q3='Q1,Q3', min='Min', max='Max'))


tab.test <- tableby(groups ~ ortho_intolerance_score + vasomotor_score + secretomotor_score + gi_score + bladder_score + pupillometer_score + total_compass_31 + a_heartburn_regurgitation + b_fullness_early_satiety + c_nausea_vomiting + d_bloating + e_upper_abdominal_pain + f_lower_abdominal_pain + total_pagi_sym_score, data=all_data %>% filter(groups != "Jack's Data"), control = mycontrols)

summary(tab.test)

	normal_SBTT_abnormal_CASS (N=24)	normal_SBTT_normal_CASS (N=30)	prolonged_SBTT (N=16)	Total (N=70)	p value
ortho_intolerance_score					0.833
Count	24	30	16	70
mean	2.417	2.533	2.125	2.400
SD	2.394	2.097	2.156	2.190
Median	2.000	3.000	2.000	2.000
Q1,Q3	0.000, 4.250	0.000, 4.000	0.000, 4.000	0.000, 4.000
Min	0.000	0.000	0.000	0.000
Max	6.000	6.000	6.000	6.000
vasomotor_score					0.786
Count	24	30	16	70
mean	0.417	0.267	0.312	0.329
SD	0.974	0.828	0.873	0.880
Median	0.000	0.000	0.000	0.000
Q1,Q3	0.000, 0.000	0.000, 0.000	0.000, 0.000	0.000, 0.000
Min	0.000	0.000	0.000	0.000
Max	3.000	3.000	3.000	3.000
secretomotor_score					0.113
Count	24	30	16	70
mean	2.125	1.433	1.312	1.643
SD	1.454	1.431	1.078	1.394
Median	2.000	1.500	1.500	2.000
Q1,Q3	1.000, 3.000	0.000, 2.000	0.000, 2.000	0.000, 3.000
Min	0.000	0.000	0.000	0.000
Max	5.000	5.000	3.000	5.000
gi_score					0.865
Count	24	30	16	70
mean	7.042	6.933	7.938	7.200
SD	5.069	4.185	4.932	4.624
Median	7.000	8.000	6.500	7.000
Q1,Q3	3.000, 11.250	3.250, 9.750	4.000, 12.000	3.000, 10.750
Min	0.000	0.000	0.000	0.000
Max	18.000	15.000	16.000	18.000
bladder_score					0.950
Count	24	30	16	70
mean	1.042	1.133	1.000	1.071
SD	1.628	2.047	1.592	1.788
Median	0.000	0.000	0.000	0.000
Q1,Q3	0.000, 1.250	0.000, 1.000	0.000, 1.500	0.000, 1.000
Min	0.000	0.000	0.000	0.000
Max	6.000	8.000	5.000	8.000
pupillometer_score					0.321
Count	24	30	16	70
mean	3.500	3.467	2.125	3.171
SD	3.867	3.371	2.802	3.435
Median	2.500	3.000	2.000	2.000
Q1,Q3	0.000, 4.500	0.250, 5.000	0.000, 3.000	0.000, 5.000
Min	0.000	0.000	0.000	0.000
Max	13.000	14.000	11.000	14.000
total_compass_31					0.724
Count	24	30	16	70
mean	16.542	15.767	14.812	15.814
SD	8.465	9.424	7.985	8.688
Median	18.000	16.500	14.000	14.500
Q1,Q3	10.500, 24.000	8.500, 19.750	10.500, 18.000	9.250, 20.750
Min	3.000	3.000	5.000	3.000
Max	30.000	38.000	37.000	38.000
a_heartburn_regurgitation					0.258
Count	24	30	16	70
mean	1.006	0.533	0.911	0.782
SD	1.081	0.777	1.234	1.011
Median	0.571	0.357	0.214	0.429
Q1,Q3	0.107, 1.464	0.000, 0.571	0.000, 1.714	0.000, 1.107
Min	0.000	0.000	0.000	0.000
Max	3.857	3.714	4.000	4.000
b_fullness_early_satiety					0.773
Count	24	30	16	70
mean	0.979	0.858	1.156	0.968
SD	0.964	0.806	1.087	0.924
Median	0.500	0.750	0.750	0.750
Q1,Q3	0.250, 1.750	0.250, 1.188	0.250, 2.062	0.250, 1.500
Min	0.000	0.000	0.000	0.000
Max	3.000	3.000	3.750	3.750
c_nausea_vomiting					0.213
Count	24	30	16	70
mean	0.500	0.344	0.500	0.433
SD	0.606	0.766	0.911	0.745
Median	0.333	0.000	0.000	0.000
Q1,Q3	0.000, 0.750	0.000, 0.250	0.000, 0.583	0.000, 0.667
Min	0.000	0.000	0.000	0.000
Max	2.333	3.333	3.000	3.333
d_bloating					0.696
Count	24	30	16	70
mean	1.417	1.533	1.188	1.414
SD	1.487	1.345	1.276	1.367
Median	1.000	1.500	0.750	1.000
Q1,Q3	0.000, 2.000	0.000, 2.500	0.000, 2.500	0.000, 2.500
Min	0.000	0.000	0.000	0.000
Max	5.000	4.500	3.000	5.000
e_upper_abdominal_pain					0.476
Count	24	30	16	70
mean	0.938	0.600	0.781	0.757
SD	1.106	1.012	1.224	1.089
Median	0.250	0.000	0.000	0.000
Q1,Q3	0.000, 2.000	0.000, 1.000	0.000, 1.250	0.000, 1.500
Min	0.000	0.000	0.000	0.000
Max	3.000	3.000	3.000	3.000
f_lower_abdominal_pain					0.488
Count	24	30	16	70
mean	1.104	0.883	0.719	0.921
SD	1.142	1.187	1.238	1.175
Median	1.000	0.000	0.000	0.000
Q1,Q3	0.000, 2.000	0.000, 2.000	0.000, 1.000	0.000, 2.000
Min	0.000	0.000	0.000	0.000
Max	3.000	4.000	4.000	4.000
total_pagi_sym_score					0.644
Count	24	30	16	70
mean	0.991	0.792	0.876	0.879
SD	0.801	0.776	0.735	0.770
Median	0.923	0.589	0.637	0.652
Q1,Q3	0.357, 1.438	0.208, 1.061	0.156, 1.543	0.208, 1.382
Min	0.000	0.000	0.000	0.000
Max	2.901	3.091	2.077	3.091

# all_data %>% filter(groups == "prolonged_SBTT")

final_spreadsheet_control <- read_excel("C:/Users/zhaoz03/OneDrive - The Mount Sinai Hospital/MY PROJECTS/Project 11. GI Pressue EVA Study/Dataset/Final spreadsheet_controls and PWH.xlsx", sheet = 1, skip = 1)

final_spreadsheet_control <- final_spreadsheet_control %>% filter(!is.na(mcass_brsa))

EVA_Controls_5min <- EVA_Controls_5min %>%
  left_join(final_spreadsheet_control %>% select(SUBJID, SBTT), by = c("combined_ID" = "SUBJID"))

EVA_Controls_5min$total_time_minue <- EVA_Controls_5min$total_time*60
EVA_Controls_5min$SBTT...34 <- as.numeric(EVA_Controls_5min$`SBTT`)

Control_Patients_5min_RMSSD <- Control_Patients_5min %>% 
  group_by(combined_ID, epoch_type) %>%
  summarize(
    mean_RMSSD_Time = mean(epoch_RMSSD_Time),
    median_RMSSD_Time = median(epoch_RMSSD_Time),
    sd_RMSSD_Time = sd(epoch_RMSSD_Time),
    mean_RMSSD_Magnitude = mean(epoch_RMSSD_Magnitude),
    median_RMSSD_Magnitude = median(epoch_RMSSD_Magnitude),
    sd_RMSSD_Magnitude = sd(epoch_RMSSD_Magnitude),
    n_epochs = n(),
    .groups = "drop"
  ) %>%
  pivot_wider(
    id_cols = combined_ID,
    names_from = epoch_type,
    values_from = c(mean_RMSSD_Time, median_RMSSD_Time, sd_RMSSD_Time, mean_RMSSD_Magnitude, median_RMSSD_Magnitude, sd_RMSSD_Magnitude, n_epochs)) %>%
  mutate(groups = "Jack's Data")

Control_Patients_5min_total_epoch <- Control_Patients_5min %>% 
  group_by(combined_ID) %>%
  summarize(
    median_RMSSD_Time_total_epoch = median(epoch_RMSSD_Time),
    median_RMSSD_Magnitude_total_epoch = median(epoch_RMSSD_Magnitude),
    .groups = "drop"
  )

Control_Patients_5min_RMSSD <- Control_Patients_5min_RMSSD %>%
  left_join(Control_Patients_5min_total_epoch)
    
EVA_Patients_5min_RMSSD <- EVA_Patients_5min %>% 
  group_by(combined_ID, groups, epoch_type) %>%
  summarize(
    mean_RMSSD_Time = mean(epoch_RMSSD_Time),
    median_RMSSD_Time = median(epoch_RMSSD_Time),
    sd_RMSSD_Time = sd(epoch_RMSSD_Time),
    mean_RMSSD_Magnitude = mean(epoch_RMSSD_Magnitude),
    median_RMSSD_Magnitude = median(epoch_RMSSD_Magnitude),
    sd_RMSSD_Magnitude = sd(epoch_RMSSD_Magnitude),
    n_epochs = n(),
    .groups = "drop"
  ) %>%
  pivot_wider(
    id_cols = c(combined_ID, groups),
    names_from = epoch_type,
    values_from = c(mean_RMSSD_Time, median_RMSSD_Time, sd_RMSSD_Time, mean_RMSSD_Magnitude, median_RMSSD_Magnitude, sd_RMSSD_Magnitude, n_epochs)
  )

EVA_Patients_5min_total_epoch <- EVA_Patients_5min %>% 
  group_by(combined_ID) %>%
  summarize(
    median_RMSSD_Time_total_epoch = median(epoch_RMSSD_Time),
    median_RMSSD_Magnitude_total_epoch = median(epoch_RMSSD_Magnitude),
    .groups = "drop"
  )

EVA_Patients_5min_RMSSD <- EVA_Patients_5min_RMSSD %>%
  left_join(EVA_Patients_5min_total_epoch)

all_data_RMSSD <- rbind(Control_Patients_5min_RMSSD, EVA_Patients_5min_RMSSD)

4.2.2 Table 1 Extension

mycontrols <- tableby.control(test = T, total = TRUE,
                              numeric.test = "kwt", cat.test = "chisq",
                              numeric.stats=c("N", "Nmiss", "mean", "sd", "median", "q1q3", "min", "max"),
                              cat.stats=c("countpct"),
                              stats.labels=list(N='Count', Nmiss='Missing Values', mean='mean', sd='SD', median='Median', q1q3='Q1,Q3', min='Min', max='Max'))


tab.test <- tableby(groups ~ total_time + total_no_missing_time + no_missing_time_ratio + total_number_of_contractions + total_contractions_frequency + sum_of_amplitude + total_number_of_contractions_over26.7 + total_contraction_rate_0 + total_contraction_rate_1_3 + total_contraction_rate_4_6 + total_contraction_rate_7_9 + total_contraction_rate_over10 + number_of_contractions_over26.7_per_hour + imputed_total_number_of_contractions + imputed_sum_of_amplitude + number_of_fast_epochs + number_of_medium_epochs + number_of_slow_epochs + number_of_total_epochs + number_of_fast_epochs_frequency + number_of_medium_epochs_frequency + number_of_slow_epochs_frequency + number_of_total_epochs_frequency + motility_index, data=all_data, control = mycontrols)

summary(tab.test)

	Jack’s Data (N=98)	normal_SBTT_abnormal_CASS (N=24)	normal_SBTT_normal_CASS (N=30)	prolonged_SBTT (N=16)	Total (N=168)	p value
total_time						< 0.001
Count	98	24	30	16	168
mean	3.762	3.848	4.415	8.778	4.369
SD	1.160	1.064	1.139	4.746	2.303
Median	3.706	3.989	4.728	7.244	3.953
Q1,Q3	2.910, 4.409	3.136, 4.507	3.754, 5.299	6.439, 8.270	3.236, 4.987
Min	1.312	1.779	1.664	5.026	1.312
Max	6.590	6.081	5.993	23.128	23.128
total_no_missing_time						0.004
Count	98	24	30	16	168
mean	3.518	3.027	3.021	4.686	3.471
SD	1.044	1.203	1.294	1.964	1.295
Median	3.428	2.870	2.875	4.985	3.343
Q1,Q3	2.870, 4.104	2.361, 3.726	1.906, 4.236	3.099, 5.907	2.600, 4.186
Min	1.307	1.107	1.074	1.010	1.010
Max	6.212	6.064	5.555	8.811	8.811
no_missing_time_ratio						< 0.001
Count	98	24	30	16	168
mean	0.942	0.777	0.706	0.638	0.847
SD	0.084	0.177	0.251	0.296	0.202
Median	0.971	0.815	0.761	0.723	0.941
Q1,Q3	0.939, 0.990	0.725, 0.894	0.601, 0.926	0.483, 0.834	0.784, 0.981
Min	0.442	0.311	0.224	0.103	0.103
Max	1.000	0.997	0.991	0.981	1.000
total_number_of_contractions						0.017
Count	98	24	30	16	168
mean	786.857	955.042	884.067	1183.812	866.048
SD	392.159	372.587	483.217	578.218	439.970
Median	734.000	941.500	724.500	1108.500	781.500
Q1,Q3	517.500, 951.750	674.000, 1086.500	545.750, 1190.500	698.500, 1522.250	553.500, 1073.000
Min	174.000	428.000	252.000	555.000	174.000
Max	2517.000	1753.000	2006.000	2582.000	2582.000
total_contractions_frequency						< 0.001
Count	98	24	30	16	168
mean	226.842	328.487	314.751	288.198	262.904
SD	91.448	76.671	193.568	153.865	127.379
Median	218.881	334.502	272.254	231.708	254.795
Q1,Q3	160.092, 272.262	280.663, 373.353	211.168, 326.458	183.643, 355.167	174.421, 328.764
Min	52.572	152.216	118.919	122.110	52.572
Max	433.827	484.767	1122.063	581.841	1122.063
sum_of_amplitude						0.009
Count	98	24	30	16	168
mean	100940.428	112627.251	103712.522	145984.258	107394.879
SD	37378.788	39934.317	48057.118	57372.612	43604.595
Median	96916.642	106655.760	87375.001	148375.073	101171.150
Q1,Q3	76487.509, 125008.950	87945.762, 136699.811	78928.684, 132581.772	105478.880, 165435.091	77920.505, 132384.909
Min	31958.930	43679.582	34593.705	60968.451	31958.930
Max	224859.417	203251.944	237168.559	262383.401	262383.401
total_number_of_contractions_over26.7						0.002
Count	98	24	30	16	168
mean	90.847	143.458	121.967	152.562	109.798
SD	68.169	74.209	109.690	95.568	83.323
Median	72.000	143.000	86.500	148.500	89.000
Q1,Q3	41.000, 127.500	103.250, 190.500	50.250, 147.000	74.000, 211.750	46.750, 151.750
Min	12.000	8.000	9.000	31.000	8.000
Max	338.000	289.000	441.000	334.000	441.000
total_contraction_rate_0						0.166
Count	98	24	30	16	168
mean	0.247	0.175	0.258	0.291	0.243
SD	0.153	0.101	0.170	0.223	0.159
Median	0.222	0.170	0.241	0.231	0.219
Q1,Q3	0.122, 0.341	0.094, 0.232	0.142, 0.307	0.135, 0.419	0.128, 0.327
Min	0.000	0.012	0.038	0.050	0.000
Max	0.685	0.389	0.869	0.970	0.970
total_contraction_rate_1_3						0.078
Count	98	24	30	16	168
mean	0.344	0.308	0.327	0.356	0.337
SD	0.093	0.056	0.095	0.105	0.091
Median	0.357	0.319	0.338	0.384	0.345
Q1,Q3	0.272, 0.411	0.263, 0.352	0.244, 0.400	0.334, 0.410	0.272, 0.401
Min	0.110	0.212	0.114	0.029	0.029
Max	0.566	0.396	0.468	0.488	0.566
total_contraction_rate_4_6						0.314
Count	98	24	30	16	168
mean	0.187	0.209	0.180	0.170	0.188
SD	0.071	0.059	0.066	0.066	0.068
Median	0.192	0.211	0.183	0.185	0.189
Q1,Q3	0.133, 0.225	0.175, 0.258	0.135, 0.213	0.128, 0.215	0.133, 0.227
Min	0.041	0.106	0.002	0.000	0.000
Max	0.364	0.318	0.366	0.251	0.366
total_contraction_rate_7_9						0.045
Count	98	24	30	16	168
mean	0.136	0.160	0.120	0.098	0.133
SD	0.082	0.069	0.070	0.060	0.077
Median	0.115	0.145	0.101	0.096	0.114
Q1,Q3	0.081, 0.192	0.109, 0.212	0.077, 0.153	0.051, 0.134	0.081, 0.185
Min	0.005	0.077	0.001	0.000	0.000
Max	0.455	0.289	0.336	0.219	0.455
total_contraction_rate_over10						0.002
Count	98	24	30	16	168
mean	0.085	0.147	0.115	0.085	0.099
SD	0.071	0.067	0.112	0.070	0.082
Median	0.067	0.151	0.078	0.060	0.077
Q1,Q3	0.030, 0.125	0.100, 0.187	0.040, 0.148	0.039, 0.124	0.038, 0.143
Min	0.000	0.025	0.002	0.000	0.000
Max	0.283	0.311	0.405	0.266	0.405
number_of_contractions_over26.7_per_hour						< 0.001
Count	98	24	30	16	168
mean	27.688	53.789	43.732	39.017	35.360
SD	23.529	38.221	35.782	32.906	30.653
Median	20.876	48.202	25.455	30.679	25.092
Q1,Q3	11.757, 37.625	29.563, 61.081	18.510, 63.387	16.338, 49.153	15.182, 46.148
Min	3.119	4.460	7.251	7.841	3.119
Max	155.521	168.976	151.467	124.414	168.976
imputed_total_number_of_contractions						< 0.001
Count	98	24	30	16	168
mean	842.918	1311.250	1286.867	2494.000	1146.345
SD	398.189	474.939	603.969	1770.949	838.864
Median	788.000	1202.500	1177.500	1795.500	968.000
Q1,Q3	577.000, 1006.000	978.750, 1659.250	782.750, 1684.750	1181.500, 3387.500	672.000, 1336.000
Min	204.000	630.000	332.000	641.000	204.000
Max	2526.000	2463.000	2731.000	6210.000	6210.000
imputed_sum_of_amplitude						< 0.001
Count	98	24	30	16	168
mean	108176.988	151095.293	153274.484	311977.985	141770.918
SD	38328.935	42423.591	59153.165	244119.520	102643.755
Median	103445.589	149617.504	140137.128	207581.029	120036.438
Q1,Q3	79719.797, 129270.590	117704.583, 177376.599	110148.167, 193260.523	169271.089, 325802.507	89592.153, 161196.714
Min	34829.877	83591.146	62203.157	112936.509	34829.877
Max	225832.269	250021.028	295903.363	950697.962	950697.962
number_of_fast_epochs						0.424
Count	98	24	30	16	168
mean	4.724	4.833	4.100	3.125	4.476
SD	4.497	3.841	4.358	2.941	4.255
Median	4.000	4.000	3.000	2.000	4.000
Q1,Q3	1.000, 7.000	2.750, 7.000	1.000, 6.000	1.000, 5.250	1.000, 7.000
Min	0.000	0.000	0.000	0.000	0.000
Max	27.000	15.000	17.000	8.000	27.000
number_of_medium_epochs						0.029
Count	98	24	30	16	168
mean	3.531	3.292	2.367	1.938	3.137
SD	2.894	3.155	3.306	1.692	2.955
Median	3.000	2.000	1.500	1.000	2.000
Q1,Q3	1.000, 5.000	1.000, 4.500	1.000, 3.000	1.000, 3.000	1.000, 5.000
Min	0.000	0.000	0.000	0.000	0.000
Max	13.000	13.000	18.000	5.000	18.000
number_of_slow_epochs						0.058
Count	98	24	30	16	168
mean	4.663	3.208	3.533	3.750	4.167
SD	3.386	3.243	3.521	3.550	3.429
Median	4.000	3.000	2.500	2.500	3.000
Q1,Q3	2.000, 6.000	1.000, 5.000	1.000, 4.000	1.000, 5.500	1.000, 6.000
Min	0.000	0.000	0.000	0.000	0.000
Max	15.000	12.000	15.000	11.000	15.000
number_of_total_epochs						0.067
Count	98	24	30	16	168
mean	12.918	11.333	10.000	8.812	11.780
SD	8.196	8.560	7.944	6.514	8.129
Median	12.000	11.500	9.000	7.000	10.000
Q1,Q3	7.000, 18.000	5.000, 16.000	5.250, 11.750	3.750, 14.000	5.750, 17.000
Min	1.000	1.000	1.000	1.000	1.000
Max	51.000	37.000	34.000	22.000	51.000
number_of_fast_epochs_frequency						0.075
Count	98	24	30	16	168
mean	1.377	1.544	1.393	0.660	1.335
SD	1.188	1.038	1.354	0.545	1.168
Median	1.108	1.310	1.131	0.524	1.107
Q1,Q3	0.378, 2.274	1.019, 1.919	0.261, 1.976	0.271, 0.990	0.381, 1.954
Min	0.000	0.000	0.000	0.000	0.000
Max	4.601	4.450	5.168	1.931	5.168
number_of_medium_epochs_frequency						0.003
Count	98	24	30	16	168
mean	1.019	0.969	0.713	0.372	0.895
SD	0.837	0.718	0.676	0.286	0.779
Median	0.847	0.886	0.604	0.369	0.735
Q1,Q3	0.339, 1.529	0.594, 1.195	0.318, 1.000	0.193, 0.524	0.313, 1.274
Min	0.000	0.000	0.000	0.000	0.000
Max	4.302	2.757	3.240	0.964	4.302
number_of_slow_epochs_frequency						0.009
Count	98	24	30	16	168
mean	1.295	0.890	1.047	0.666	1.133
SD	0.850	0.790	0.751	0.511	0.820
Median	1.214	0.779	0.924	0.611	1.005
Q1,Q3	0.689, 1.832	0.358, 1.332	0.426, 1.661	0.334, 0.915	0.506, 1.650
Min	0.000	0.000	0.000	0.000	0.000
Max	3.648	3.232	2.700	1.572	3.648
number_of_total_epochs_frequency						< 0.001
Count	98	24	30	16	168
mean	3.691	3.402	3.153	1.697	3.364
SD	2.027	1.946	1.848	0.828	1.975
Median	3.430	3.144	3.081	1.513	3.096
Q1,Q3	2.159, 4.985	2.119, 4.317	1.765, 3.962	1.136, 2.300	1.778, 4.430
Min	0.302	0.552	0.384	0.326	0.302
Max	9.386	7.846	7.753	3.143	9.386
motility_index						0.015
Count	98	24	30	16	168
mean	18.001	18.355	18.079	18.782	18.140
SD	0.873	0.758	1.022	0.872	0.911
Median	18.086	18.500	17.924	19.016	18.151
Q1,Q3	17.521, 18.615	17.857, 18.820	17.565, 18.899	18.020, 19.289	17.584, 18.753
Min	15.750	16.816	15.981	17.360	15.750
Max	20.154	19.678	19.826	20.278	20.278

mycontrols <- tableby.control(test = T, total = TRUE,
                              numeric.test = "kwt", cat.test = "chisq",
                              numeric.stats=c("N", "Nmiss", "mean", "sd", "median", "q1q3", "min", "max"),
                              cat.stats=c("countpct"),
                              stats.labels=list(N='Count', Nmiss='Missing Values', mean='mean', sd='SD', median='Median', q1q3='Q1,Q3', min='Min', max='Max'))


tab.test <- tableby(groups ~ `median_RMSSD_Time_fast epoch` + `median_RMSSD_Time_medium epoch` + `median_RMSSD_Time_slow epoch`+ median_RMSSD_Time_total_epoch + `median_RMSSD_Magnitude_fast epoch` + `median_RMSSD_Magnitude_medium epoch` + `median_RMSSD_Magnitude_slow epoch` + median_RMSSD_Magnitude_total_epoch + `n_epochs_fast epoch` + `n_epochs_medium epoch` + `n_epochs_slow epoch`, data=all_data_RMSSD, control = mycontrols)

summary(tab.test)

	Jack’s Data (N=98)	normal_SBTT_abnormal_CASS (N=24)	normal_SBTT_normal_CASS (N=30)	prolonged_SBTT (N=16)	Total (N=168)	p value
median_RMSSD_Time_fast epoch						0.110
Count	83	22	24	14	143
Missing Values	15	2	6	2	25
mean	50.465	52.652	52.751	54.094	51.541
SD	4.285	6.043	6.671	7.158	5.446
Median	49.280	51.373	52.530	53.617	50.190
Q1,Q3	47.425, 53.165	48.320, 56.224	47.042, 57.315	48.523, 57.059	47.672, 55.133
Min	43.860	45.105	41.540	43.885	41.540
Max	63.850	71.460	68.990	71.790	71.790
median_RMSSD_Time_medium epoch						0.018
Count	87	20	24	13	144
Missing Values	11	4	6	3	24
mean	63.418	69.505	65.233	67.195	64.907
SD	7.848	11.156	7.523	11.727	8.889
Median	61.950	68.787	64.795	64.430	63.220
Q1,Q3	58.100, 65.892	64.308, 71.417	60.700, 69.285	60.540, 68.960	59.020, 68.398
Min	52.110	54.780	53.570	50.480	50.480
Max	96.330	101.345	80.560	95.890	101.345
median_RMSSD_Time_slow epoch						0.224
Count	90	19	26	13	148
Missing Values	8	5	4	3	20
mean	78.458	84.218	84.863	79.166	80.385
SD	9.098	15.395	14.989	7.530	11.394
Median	77.975	80.190	82.920	77.400	79.388
Q1,Q3	71.345, 84.540	76.070, 88.485	73.539, 95.785	73.420, 84.980	72.803, 86.286
Min	52.200	64.420	64.010	68.985	52.200
Max	101.360	138.230	117.070	91.350	138.230
median_RMSSD_Time_total_epoch						0.764
Count	98	24	30	16	168
mean	64.060	65.017	62.848	65.641	64.131
SD	10.438	17.527	9.990	10.044	11.516
Median	62.395	63.505	62.873	66.955	63.110
Q1,Q3	57.683, 70.236	55.435, 68.371	58.581, 68.760	60.663, 69.500	57.415, 69.545
Min	46.620	46.790	46.380	47.840	46.380
Max	98.730	138.230	83.040	88.090	138.230
median_RMSSD_Magnitude_fast epoch						0.368
Count	83	22	24	14	143
Missing Values	15	2	6	2	25
mean	65.556	73.100	72.270	73.601	68.631
SD	24.956	29.681	28.263	28.109	26.551
Median	61.095	72.650	68.257	75.285	66.490
Q1,Q3	47.792, 75.675	53.935, 84.433	52.824, 87.936	62.897, 79.865	49.400, 82.235
Min	20.160	24.450	30.250	26.515	20.160
Max	157.550	158.900	130.970	131.850	158.900
median_RMSSD_Magnitude_medium epoch						0.071
Count	87	20	24	13	144
Missing Values	11	4	6	3	24
mean	56.492	70.329	55.683	52.408	57.910
SD	23.552	29.318	20.858	21.607	24.152
Median	49.800	63.540	47.740	47.980	51.140
Q1,Q3	40.300, 63.960	50.250, 81.223	40.731, 73.131	36.505, 55.990	40.835, 67.465
Min	28.345	29.715	27.860	30.220	27.860
Max	149.150	151.905	98.165	112.990	151.905
median_RMSSD_Magnitude_slow epoch						0.117
Count	90	19	26	13	148
Missing Values	8	5	4	3	20
mean	43.154	45.317	42.074	58.182	44.562
SD	26.142	16.941	18.883	39.329	25.585
Median	35.885	44.630	42.317	45.160	40.722
Q1,Q3	26.372, 49.825	34.177, 48.555	29.869, 48.213	38.030, 50.540	28.337, 49.554
Min	20.100	22.930	17.460	28.085	17.460
Max	180.510	87.080	111.320	177.435	180.510
median_RMSSD_Magnitude_total_epoch						0.054
Count	98	24	30	16	168
mean	52.018	63.340	59.000	55.252	55.190
SD	21.697	24.650	26.846	17.715	22.990
Median	47.945	57.235	52.045	55.907	51.043
Q1,Q3	37.714, 59.303	49.212, 75.714	42.013, 66.657	39.045, 67.170	39.865, 62.062
Min	21.700	24.450	22.125	31.570	21.700
Max	136.030	139.395	130.970	93.415	139.395
n_epochs_fast epoch						0.306
Count	83	22	24	14	143
Missing Values	15	2	6	2	25
mean	5.578	5.273	5.125	3.571	5.259
SD	4.370	3.706	4.297	2.875	4.141
Median	5.000	4.500	4.500	2.000	5.000
Q1,Q3	2.500, 7.000	3.000, 7.000	1.750, 6.250	1.000, 5.750	2.000, 7.000
Min	1.000	1.000	1.000	1.000	1.000
Max	27.000	15.000	17.000	8.000	27.000
n_epochs_medium epoch						0.042
Count	87	20	24	13	144
Missing Values	11	4	6	3	24
mean	3.977	3.950	2.958	2.385	3.660
SD	2.766	3.052	3.458	1.557	2.875
Median	3.000	3.000	2.000	2.000	3.000
Q1,Q3	2.000, 6.000	2.000, 6.000	1.000, 3.000	1.000, 3.000	1.750, 5.000
Min	1.000	1.000	1.000	1.000	1.000
Max	13.000	13.000	18.000	5.000	18.000
n_epochs_slow epoch						0.196
Count	90	19	26	13	148
Missing Values	8	5	4	3	20
mean	5.078	4.053	4.077	4.615	4.730
SD	3.219	3.135	3.475	3.380	3.267
Median	4.000	3.000	3.000	4.000	4.000
Q1,Q3	3.000, 7.000	1.500, 5.000	1.250, 5.500	2.000, 7.000	2.000, 6.250
Min	1.000	1.000	1.000	1.000	1.000
Max	15.000	12.000	15.000	11.000	15.000

4.2.3 Post-hoc Pairwise Comparisons across 4 Groups

mycontrols <- tableby.control(test = T, total = TRUE,
                              numeric.test = "kwt", cat.test = "chisq",
                              numeric.stats=c("N", "Nmiss", "mean", "sd", "median", "q1q3", "min", "max"),
                              cat.stats=c("countpct"),
                              stats.labels=list(N='Count', Nmiss='Missing Values', mean='mean', sd='SD', median='Median', q1q3='Q1,Q3', min='Min', max='Max'))


tab.test <- tableby(groups ~ total_contractions_frequency + total_contraction_rate_over10 + number_of_contractions_over26.7_per_hour + number_of_total_epochs_frequency + imputed_sum_of_amplitude + motility_index, data=all_data, control = mycontrols)

summary(tab.test)

	Jack’s Data (N=98)	normal_SBTT_abnormal_CASS (N=24)	normal_SBTT_normal_CASS (N=30)	prolonged_SBTT (N=16)	Total (N=168)	p value
total_contractions_frequency						< 0.001
Count	98	24	30	16	168
mean	226.842	328.487	314.751	288.198	262.904
SD	91.448	76.671	193.568	153.865	127.379
Median	218.881	334.502	272.254	231.708	254.795
Q1,Q3	160.092, 272.262	280.663, 373.353	211.168, 326.458	183.643, 355.167	174.421, 328.764
Min	52.572	152.216	118.919	122.110	52.572
Max	433.827	484.767	1122.063	581.841	1122.063
total_contraction_rate_over10						0.002
Count	98	24	30	16	168
mean	0.085	0.147	0.115	0.085	0.099
SD	0.071	0.067	0.112	0.070	0.082
Median	0.067	0.151	0.078	0.060	0.077
Q1,Q3	0.030, 0.125	0.100, 0.187	0.040, 0.148	0.039, 0.124	0.038, 0.143
Min	0.000	0.025	0.002	0.000	0.000
Max	0.283	0.311	0.405	0.266	0.405
number_of_contractions_over26.7_per_hour						< 0.001
Count	98	24	30	16	168
mean	27.688	53.789	43.732	39.017	35.360
SD	23.529	38.221	35.782	32.906	30.653
Median	20.876	48.202	25.455	30.679	25.092
Q1,Q3	11.757, 37.625	29.563, 61.081	18.510, 63.387	16.338, 49.153	15.182, 46.148
Min	3.119	4.460	7.251	7.841	3.119
Max	155.521	168.976	151.467	124.414	168.976
number_of_total_epochs_frequency						< 0.001
Count	98	24	30	16	168
mean	3.691	3.402	3.153	1.697	3.364
SD	2.027	1.946	1.848	0.828	1.975
Median	3.430	3.144	3.081	1.513	3.096
Q1,Q3	2.159, 4.985	2.119, 4.317	1.765, 3.962	1.136, 2.300	1.778, 4.430
Min	0.302	0.552	0.384	0.326	0.302
Max	9.386	7.846	7.753	3.143	9.386
imputed_sum_of_amplitude						< 0.001
Count	98	24	30	16	168
mean	108176.988	151095.293	153274.484	311977.985	141770.918
SD	38328.935	42423.591	59153.165	244119.520	102643.755
Median	103445.589	149617.504	140137.128	207581.029	120036.438
Q1,Q3	79719.797, 129270.590	117704.583, 177376.599	110148.167, 193260.523	169271.089, 325802.507	89592.153, 161196.714
Min	34829.877	83591.146	62203.157	112936.509	34829.877
Max	225832.269	250021.028	295903.363	950697.962	950697.962
motility_index						0.015
Count	98	24	30	16	168
mean	18.001	18.355	18.079	18.782	18.140
SD	0.873	0.758	1.022	0.872	0.911
Median	18.086	18.500	17.924	19.016	18.151
Q1,Q3	17.521, 18.615	17.857, 18.820	17.565, 18.899	18.020, 19.289	17.584, 18.753
Min	15.750	16.816	15.981	17.360	15.750
Max	20.154	19.678	19.826	20.278	20.278

mycontrols <- tableby.control(test = T, total = TRUE,
                              numeric.test = "kwt", cat.test = "chisq",
                              numeric.stats=c("N", "Nmiss", "mean", "sd", "median", "q1q3", "min", "max"),
                              cat.stats=c("countpct"),
                              stats.labels=list(N='Count', Nmiss='Missing Values', mean='mean', sd='SD', median='Median', q1q3='Q1,Q3', min='Min', max='Max'))


tab.test <- tableby(groups ~ `median_RMSSD_Time_fast epoch` + `median_RMSSD_Time_medium epoch` + `median_RMSSD_Time_slow epoch`+ median_RMSSD_Time_total_epoch + `median_RMSSD_Magnitude_fast epoch` + `median_RMSSD_Magnitude_medium epoch` + `median_RMSSD_Magnitude_slow epoch` + median_RMSSD_Magnitude_total_epoch + `n_epochs_fast epoch` + `n_epochs_medium epoch` + `n_epochs_slow epoch`, data=all_data_RMSSD, control = mycontrols)

summary(tab.test)

	Jack’s Data (N=98)	normal_SBTT_abnormal_CASS (N=24)	normal_SBTT_normal_CASS (N=30)	prolonged_SBTT (N=16)	Total (N=168)	p value
median_RMSSD_Time_fast epoch						0.110
Count	83	22	24	14	143
Missing Values	15	2	6	2	25
mean	50.465	52.652	52.751	54.094	51.541
SD	4.285	6.043	6.671	7.158	5.446
Median	49.280	51.373	52.530	53.617	50.190
Q1,Q3	47.425, 53.165	48.320, 56.224	47.042, 57.315	48.523, 57.059	47.672, 55.133
Min	43.860	45.105	41.540	43.885	41.540
Max	63.850	71.460	68.990	71.790	71.790
median_RMSSD_Time_medium epoch						0.018
Count	87	20	24	13	144
Missing Values	11	4	6	3	24
mean	63.418	69.505	65.233	67.195	64.907
SD	7.848	11.156	7.523	11.727	8.889
Median	61.950	68.787	64.795	64.430	63.220
Q1,Q3	58.100, 65.892	64.308, 71.417	60.700, 69.285	60.540, 68.960	59.020, 68.398
Min	52.110	54.780	53.570	50.480	50.480
Max	96.330	101.345	80.560	95.890	101.345
median_RMSSD_Time_slow epoch						0.224
Count	90	19	26	13	148
Missing Values	8	5	4	3	20
mean	78.458	84.218	84.863	79.166	80.385
SD	9.098	15.395	14.989	7.530	11.394
Median	77.975	80.190	82.920	77.400	79.388
Q1,Q3	71.345, 84.540	76.070, 88.485	73.539, 95.785	73.420, 84.980	72.803, 86.286
Min	52.200	64.420	64.010	68.985	52.200
Max	101.360	138.230	117.070	91.350	138.230
median_RMSSD_Time_total_epoch						0.764
Count	98	24	30	16	168
mean	64.060	65.017	62.848	65.641	64.131
SD	10.438	17.527	9.990	10.044	11.516
Median	62.395	63.505	62.873	66.955	63.110
Q1,Q3	57.683, 70.236	55.435, 68.371	58.581, 68.760	60.663, 69.500	57.415, 69.545
Min	46.620	46.790	46.380	47.840	46.380
Max	98.730	138.230	83.040	88.090	138.230
median_RMSSD_Magnitude_fast epoch						0.368
Count	83	22	24	14	143
Missing Values	15	2	6	2	25
mean	65.556	73.100	72.270	73.601	68.631
SD	24.956	29.681	28.263	28.109	26.551
Median	61.095	72.650	68.257	75.285	66.490
Q1,Q3	47.792, 75.675	53.935, 84.433	52.824, 87.936	62.897, 79.865	49.400, 82.235
Min	20.160	24.450	30.250	26.515	20.160
Max	157.550	158.900	130.970	131.850	158.900
median_RMSSD_Magnitude_medium epoch						0.071
Count	87	20	24	13	144
Missing Values	11	4	6	3	24
mean	56.492	70.329	55.683	52.408	57.910
SD	23.552	29.318	20.858	21.607	24.152
Median	49.800	63.540	47.740	47.980	51.140
Q1,Q3	40.300, 63.960	50.250, 81.223	40.731, 73.131	36.505, 55.990	40.835, 67.465
Min	28.345	29.715	27.860	30.220	27.860
Max	149.150	151.905	98.165	112.990	151.905
median_RMSSD_Magnitude_slow epoch						0.117
Count	90	19	26	13	148
Missing Values	8	5	4	3	20
mean	43.154	45.317	42.074	58.182	44.562
SD	26.142	16.941	18.883	39.329	25.585
Median	35.885	44.630	42.317	45.160	40.722
Q1,Q3	26.372, 49.825	34.177, 48.555	29.869, 48.213	38.030, 50.540	28.337, 49.554
Min	20.100	22.930	17.460	28.085	17.460
Max	180.510	87.080	111.320	177.435	180.510
median_RMSSD_Magnitude_total_epoch						0.054
Count	98	24	30	16	168
mean	52.018	63.340	59.000	55.252	55.190
SD	21.697	24.650	26.846	17.715	22.990
Median	47.945	57.235	52.045	55.907	51.043
Q1,Q3	37.714, 59.303	49.212, 75.714	42.013, 66.657	39.045, 67.170	39.865, 62.062
Min	21.700	24.450	22.125	31.570	21.700
Max	136.030	139.395	130.970	93.415	139.395
n_epochs_fast epoch						0.306
Count	83	22	24	14	143
Missing Values	15	2	6	2	25
mean	5.578	5.273	5.125	3.571	5.259
SD	4.370	3.706	4.297	2.875	4.141
Median	5.000	4.500	4.500	2.000	5.000
Q1,Q3	2.500, 7.000	3.000, 7.000	1.750, 6.250	1.000, 5.750	2.000, 7.000
Min	1.000	1.000	1.000	1.000	1.000
Max	27.000	15.000	17.000	8.000	27.000
n_epochs_medium epoch						0.042
Count	87	20	24	13	144
Missing Values	11	4	6	3	24
mean	3.977	3.950	2.958	2.385	3.660
SD	2.766	3.052	3.458	1.557	2.875
Median	3.000	3.000	2.000	2.000	3.000
Q1,Q3	2.000, 6.000	2.000, 6.000	1.000, 3.000	1.000, 3.000	1.750, 5.000
Min	1.000	1.000	1.000	1.000	1.000
Max	13.000	13.000	18.000	5.000	18.000
n_epochs_slow epoch						0.196
Count	90	19	26	13	148
Missing Values	8	5	4	3	20
mean	5.078	4.053	4.077	4.615	4.730
SD	3.219	3.135	3.475	3.380	3.267
Median	4.000	3.000	3.000	4.000	4.000
Q1,Q3	3.000, 7.000	1.500, 5.000	1.250, 5.500	2.000, 7.000	2.000, 6.250
Min	1.000	1.000	1.000	1.000	1.000
Max	15.000	12.000	15.000	11.000	15.000

library(FSA)

combined_results <- data.frame()

for (var in c("total_contractions_frequency", "total_contraction_rate_over10", "number_of_contractions_over26.7_per_hour", "number_of_total_epochs_frequency", "imputed_sum_of_amplitude", "motility_index")) {
  # Perform Dunn's test
  dunn <- dunnTest(as.formula(paste(var, "~ groups")), data = all_data, method = "bonferroni")
  
  # Extract and format results
  dunn_df <- dunn$res %>%
    mutate(Variable = var) %>%
    select(Variable, Comparison, P.adj) %>%
    rename(P.adjusted = P.adj)
  
  # Combine into one table
  combined_results <- bind_rows(combined_results, dunn_df)
}

# View the combined table
combined_results <- combined_results %>%
  mutate(P.adjusted = format(round(P.adjusted, 3), nsmall = 3)) %>%
  select(Variable, Comparison, P.adjusted) %>%
  pivot_wider(names_from = Comparison, values_from = P.adjusted) 

combined_results <- data.frame()

for (var in c("total_contractions_frequency", "total_contraction_rate_over10", "number_of_contractions_over26.7_per_hour", "number_of_total_epochs_frequency", "imputed_sum_of_amplitude", "motility_index")) {
  # Perform Dunn's test
  dunn <- dunnTest(as.formula(paste(var, "~ groups")), data = all_data, method = "bonferroni")
  
  # Extract and format results
  dunn_df <- dunn$res %>%
    mutate(response = var) %>%
    select(response, Comparison, P.adj) %>%
    rename(P.adjusted = P.adj)
  
  # Combine into one table
  combined_results <- bind_rows(combined_results, dunn_df)
}

# View the combined table
# combined_results
knitr::kable(combined_results, format = "html")

response	Comparison	P.adjusted
total_contractions_frequency	Jack’s Data - normal_SBTT_abnormal_CASS	0.0000310
total_contractions_frequency	Jack’s Data - normal_SBTT_normal_CASS	0.0597706
total_contractions_frequency	normal_SBTT_abnormal_CASS - normal_SBTT_normal_CASS	0.4064831
total_contractions_frequency	Jack’s Data - prolonged_SBTT	1.0000000
total_contractions_frequency	normal_SBTT_abnormal_CASS - prolonged_SBTT	0.1932298
total_contractions_frequency	normal_SBTT_normal_CASS - prolonged_SBTT	1.0000000
total_contraction_rate_over10	Jack’s Data - normal_SBTT_abnormal_CASS	0.0006668
total_contraction_rate_over10	Jack’s Data - normal_SBTT_normal_CASS	1.0000000
total_contraction_rate_over10	normal_SBTT_abnormal_CASS - normal_SBTT_normal_CASS	0.0923847
total_contraction_rate_over10	Jack’s Data - prolonged_SBTT	1.0000000
total_contraction_rate_over10	normal_SBTT_abnormal_CASS - prolonged_SBTT	0.0476134
total_contraction_rate_over10	normal_SBTT_normal_CASS - prolonged_SBTT	1.0000000
number_of_contractions_over26.7_per_hour	Jack’s Data - normal_SBTT_abnormal_CASS	0.0008870
number_of_contractions_over26.7_per_hour	Jack’s Data - normal_SBTT_normal_CASS	0.0842242
number_of_contractions_over26.7_per_hour	normal_SBTT_abnormal_CASS - normal_SBTT_normal_CASS	1.0000000
number_of_contractions_over26.7_per_hour	Jack’s Data - prolonged_SBTT	1.0000000
number_of_contractions_over26.7_per_hour	normal_SBTT_abnormal_CASS - prolonged_SBTT	0.7461671
number_of_contractions_over26.7_per_hour	normal_SBTT_normal_CASS - prolonged_SBTT	1.0000000
number_of_total_epochs_frequency	Jack’s Data - normal_SBTT_abnormal_CASS	1.0000000
number_of_total_epochs_frequency	Jack’s Data - normal_SBTT_normal_CASS	1.0000000
number_of_total_epochs_frequency	normal_SBTT_abnormal_CASS - normal_SBTT_normal_CASS	1.0000000
number_of_total_epochs_frequency	Jack’s Data - prolonged_SBTT	0.0003179
number_of_total_epochs_frequency	normal_SBTT_abnormal_CASS - prolonged_SBTT	0.0189014
number_of_total_epochs_frequency	normal_SBTT_normal_CASS - prolonged_SBTT	0.0475040
imputed_sum_of_amplitude	Jack’s Data - normal_SBTT_abnormal_CASS	0.0004672
imputed_sum_of_amplitude	Jack’s Data - normal_SBTT_normal_CASS	0.0007496
imputed_sum_of_amplitude	normal_SBTT_abnormal_CASS - normal_SBTT_normal_CASS	1.0000000
imputed_sum_of_amplitude	Jack’s Data - prolonged_SBTT	0.0000000
imputed_sum_of_amplitude	normal_SBTT_abnormal_CASS - prolonged_SBTT	0.1850503
imputed_sum_of_amplitude	normal_SBTT_normal_CASS - prolonged_SBTT	0.0606662
motility_index	Jack’s Data - normal_SBTT_abnormal_CASS	0.5064354
motility_index	Jack’s Data - normal_SBTT_normal_CASS	1.0000000
motility_index	normal_SBTT_abnormal_CASS - normal_SBTT_normal_CASS	1.0000000
motility_index	Jack’s Data - prolonged_SBTT	0.0174224
motility_index	normal_SBTT_abnormal_CASS - prolonged_SBTT	1.0000000
motility_index	normal_SBTT_normal_CASS - prolonged_SBTT	0.1456734

4.2.4 Bar Plot

library(ggsignif)
vars <- c("total_contractions_frequency", "total_contraction_rate_over10", "number_of_contractions_over26.7_per_hour", "number_of_total_epochs_frequency", "imputed_sum_of_amplitude", "motility_index")

medians <- all_data %>%
  filter(!is.na(groups)) %>%
  group_by(groups) %>%
  summarise(across(all_of(vars), ~ median(.x, na.rm = TRUE)), .groups = "drop") %>%
  tidyr::pivot_longer(-groups, names_to = "response", values_to = "median")

combined_results <- combined_results %>% 
  # filter(P.adjusted < 0.05) %>%
  separate(Comparison, into = c("group1","group2"), sep = "\\s*-\\s*")

dunn_sig <- combined_results %>%
  filter(P.adjusted < 0.05) %>%
  mutate(
    label = ifelse(P.adjusted < 0.001, "p<0.001", sprintf("p=%.3f", P.adjusted)),
    label_sig = case_when(
      P.adjusted < 0.0001 ~ "****",
      P.adjusted < 0.001  ~ "***",
      P.adjusted < 0.01   ~ "**",
      P.adjusted < 0.05   ~ "*",
      TRUE ~ "ns"
  )) %>%
  select(response, group1, group2, label, label_sig)

y_stats <- medians %>%
  group_by(response) %>%
  summarise(
    y_max = max(median, na.rm = TRUE),
    y_min = min(median, na.rm = TRUE),
    y_rng = y_max - y_min,
    .groups = "drop"
  ) %>%
  mutate(
    pad  = ifelse(y_rng == 0, 0.10 * pmax(abs(y_max), 1), 0.08 * y_rng),
    step = ifelse(y_rng == 0, 0.08 * pmax(abs(y_max), 1), 0.10 * y_rng)
  )

anno <- dunn_sig %>%
  left_join(y_stats, by = "response") %>%
  group_by(response) %>%
  arrange(group1, group2, .by_group = TRUE) %>%
  mutate(y_position = y_max + pad + (row_number() - 1) * step) %>%
  ungroup()

anno_plot <- anno %>%
  transmute(
    response,
    xmin = group1,
    xmax = group2,
    annotations = label,
    annotations_sig = label_sig,
    y_position = y_position
  )

# Align factor levels so bars and brackets share the same x categories
grp_levels <- sort(unique(medians$groups))
medians$groups <- factor(medians$groups, levels = grp_levels)
anno_plot$xmin <- factor(anno_plot$xmin, levels = grp_levels)
anno_plot$xmax <- factor(anno_plot$xmax, levels = grp_levels)

ggplot(medians %>% 
          mutate(
            groups = case_when(
              groups == "Jack's Data" ~ "Healthy Controls",
              groups == "normal_SBTT_normal_CASS" ~ "Normal PWH",
              groups == "normal_SBTT_abnormal_CASS" ~ "PWH with AN",
              groups == "prolonged_SBTT" ~ "PWH with Prolonged SBTT"),
            response = case_when(
              response == 'total_contractions_frequency' ~ 'Contractions per hour',
              response == 'total_contraction_rate_over10' ~ 'Fraction of time spent rapidly contracting',
              response == 'number_of_contractions_over26.7_per_hour' ~ 'High amplitude contractions per hour',
              response == 'number_of_total_epochs_frequency' ~ 'Rhythmic intervals per hour',
              response == 'imputed_sum_of_amplitude' ~ 'Sum of the amplitudes',
              response == 'motility_index' ~ 'Motility index',
            )
          ), 
       aes(x = groups, y = median, fill = groups)) +
  geom_col(width = 0.7) +
  geom_signif(
    data = anno_plot %>% 
          mutate(
            xmin = case_when(
              xmin == "Jack's Data" ~ "Healthy Controls",
              xmin == "normal_SBTT_normal_CASS" ~ "Normal PWH",
              xmin == "normal_SBTT_abnormal_CASS" ~ "PWH with AN",
              xmin == "prolonged_SBTT" ~ "PWH with Prolonged SBTT"),
            xmax = case_when(
              xmax == "Jack's Data" ~ "Healthy Controls",
              xmax == "normal_SBTT_normal_CASS" ~ "Normal PWH",
              xmax == "normal_SBTT_abnormal_CASS" ~ "PWH with AN",
              xmax == "prolonged_SBTT" ~ "PWH with Prolonged SBTT"),            
            response = case_when(
              response == 'total_contractions_frequency' ~ 'Contractions per hour',
              response == 'total_contraction_rate_over10' ~ 'Fraction of time spent rapidly contracting',
              response == 'number_of_contractions_over26.7_per_hour' ~ 'High amplitude contractions per hour',
              response == 'number_of_total_epochs_frequency' ~ 'Rhythmic intervals per hour',
              response == 'imputed_sum_of_amplitude' ~ 'Sum of the amplitudes',
              response == 'motility_index' ~ 'Motility index',
            )
          ),
    manual = TRUE,            # REQUIRED for supplying data columns
    inherit.aes = FALSE,      # don't inherit fill=group, etc.
    aes(xmin = xmin, xmax = xmax,
        annotations = annotations,
        y_position = y_position),
    tip_length = 0.01,
    textsize = 3
  ) +
  facet_wrap(~ response, scales = "free_y") +
  scale_fill_brewer(palette = "Set2", name = "Group") +
  labs(x = NULL, y = "Median", title = "Medians with Dunn (Bonferroni)") +
  theme_minimal(base_size = 12) +
  theme(
    strip.text = element_text(size = 13, face = "bold"),
    panel.grid.minor = element_blank(),
    axis.text.x = element_text(angle = 20, hjust = 1, face = "bold"),
    axis.text.y = element_text(face = "bold")
  )

ggplot(medians %>% 
          mutate(
            groups = case_when(
              groups == "Jack's Data" ~ "Healthy Controls",
              groups == "normal_SBTT_normal_CASS" ~ "Normal PWH",
              groups == "normal_SBTT_abnormal_CASS" ~ "PWH with AN",
              groups == "prolonged_SBTT" ~ "PWH with Prolonged SBTT"),
            response = case_when(
              response == 'total_contractions_frequency' ~ 'Contractions per hour',
              response == 'total_contraction_rate_over10' ~ 'Fraction of time spent rapidly contracting',
              response == 'number_of_contractions_over26.7_per_hour' ~ 'High amplitude contractions per hour',
              response == 'number_of_total_epochs_frequency' ~ 'Rhythmic intervals per hour',
              response == 'imputed_sum_of_amplitude' ~ 'Sum of the amplitudes',
              response == 'motility_index' ~ 'Motility index',
            )
          ), 
       aes(x = groups, y = median, fill = groups)) +
  geom_col(width = 0.7) +
  geom_signif(
    data = anno_plot %>% 
          mutate(
            xmin = case_when(
              xmin == "Jack's Data" ~ "Healthy Controls",
              xmin == "normal_SBTT_normal_CASS" ~ "Normal PWH",
              xmin == "normal_SBTT_abnormal_CASS" ~ "PWH with AN",
              xmin == "prolonged_SBTT" ~ "PWH with Prolonged SBTT"),
            xmax = case_when(
              xmax == "Jack's Data" ~ "Healthy Controls",
              xmax == "normal_SBTT_normal_CASS" ~ "Normal PWH",
              xmax == "normal_SBTT_abnormal_CASS" ~ "PWH with AN",
              xmax == "prolonged_SBTT" ~ "PWH with Prolonged SBTT"),            
            response = case_when(
              response == 'total_contractions_frequency' ~ 'Contractions per hour',
              response == 'total_contraction_rate_over10' ~ 'Fraction of time spent rapidly contracting',
              response == 'number_of_contractions_over26.7_per_hour' ~ 'High amplitude contractions per hour',
              response == 'number_of_total_epochs_frequency' ~ 'Rhythmic intervals per hour',
              response == 'imputed_sum_of_amplitude' ~ 'Sum of the amplitudes',
              response == 'motility_index' ~ 'Motility index',
            )
          ),
    manual = TRUE,            # REQUIRED for supplying data columns
    inherit.aes = FALSE,      # don't inherit fill=group, etc.
    aes(xmin = xmin, xmax = xmax,
        annotations = annotations_sig,
        y_position = y_position),
    tip_length = 0.01,
    textsize = 3
  ) +
  facet_wrap(~ response, scales = "free_y") +
  scale_fill_brewer(palette = "Set2", name = "Group") +
  labs(x = NULL, y = "Median", title = "Medians with Dunn (Bonferroni)") +
  theme_minimal(base_size = 12) +
  theme(
    strip.text = element_text(size = 13, face = "bold"),
    panel.grid.minor = element_blank(),
    axis.text.x = element_text(angle = 20, hjust = 1, face = "bold"),
    axis.text.y = element_text(face = "bold")
  )

4.3 Table 3 - Removed

EVA_Controls_5min <- EVA_Controls_5min %>%
  left_join(Control_data_eva, by = c("combined_ID" = "SUBJID"))
EVA_Patients_5min <- EVA_Patients_5min %>%
  left_join(PWH_data, by = c("combined_ID" = "ID"))
Control_Patients_5min <- Control_Patients_5min %>%
  left_join(Control_data_jack, by = c("combined_ID" = "combined_ID"))

EVA_Controls_5min$source <- "EVA Controls"

all_data <- rbind(Control_Patients_5min, EVA_Patients_5min) #, EVA_Controls_5min)
all_data <- all_data %>% group_by(combined_ID) %>% slice(1)

mycontrols <- tableby.control(test = T, total = TRUE,
                              numeric.test = "kwt", cat.test = "chisq",
                              numeric.stats=c("N", "Nmiss", "mean", "sd", "median", "q1q3", "min", "max"),
                              cat.stats=c("countpct"),
                              stats.labels=list(N='Count', Nmiss='Missing Values', mean='mean', sd='SD', median='Median', q1q3='Q1,Q3', min='Min', max='Max'))


tab.test <- tableby(source ~ total_time + total_no_missing_time + no_missing_time_ratio + total_number_of_contractions + sum_of_amplitude + total_number_of_contractions_over26.7 + total_contraction_rate_0 + total_contraction_rate_1_3 + total_contraction_rate_4_6 + total_contraction_rate_7_9 + total_contraction_rate_over10 + number_of_contractions_over26.7_per_hour + imputed_total_number_of_contractions + imputed_sum_of_amplitude + number_of_fast_epochs + number_of_medium_epochs + number_of_slow_epochs + number_of_total_epochs + number_of_fast_epochs_frequency + number_of_medium_epochs_frequency + number_of_slow_epochs_frequency + number_of_total_epochs_frequency + motility_index, data=all_data, control = mycontrols)

summary(tab.test)

	EVA Patient (N=71)	Jack’s Data (N=98)	Total (N=169)	p value
total_time				< 0.001
Count	71	98	169
mean	5.191	3.762	4.363
SD	3.099	1.160	2.297
Median	4.703	3.706	3.951
Q1,Q3	3.716, 5.750	2.910, 4.409	3.245, 4.973
Min	1.664	1.312	1.312
Max	23.128	6.590	23.128
total_no_missing_time				0.226
Count	71	98	169
mean	3.393	3.518	3.466
SD	1.579	1.044	1.293
Median	3.063	3.428	3.338
Q1,Q3	2.425, 4.372	2.870, 4.104	2.603, 4.170
Min	1.010	1.307	1.010
Max	8.811	6.212	8.811
no_missing_time_ratio				< 0.001
Count	71	98	169
mean	0.716	0.942	0.847
SD	0.241	0.084	0.202
Median	0.766	0.971	0.941
Q1,Q3	0.613, 0.897	0.939, 0.990	0.788, 0.981
Min	0.103	0.442	0.103
Max	0.997	1.000	1.000
total_number_of_contractions				0.012
Count	71	98	169
mean	973.676	786.857	865.343
SD	477.890	392.159	438.755
Median	885.000	734.000	780.000
Q1,Q3	615.000, 1239.500	517.500, 951.750	554.000, 1073.000
Min	252.000	174.000	174.000
Max	2582.000	2517.000	2582.000
sum_of_amplitude				0.051
Count	71	98	169
mean	115979.438	100940.428	107258.592
SD	49744.617	37378.788	43510.713
Median	106323.294	96916.642	101003.974
Q1,Q3	82820.367, 148375.073	76487.509, 125008.950	77963.451, 132296.825
Min	34593.705	31958.930	31958.930
Max	262383.401	224859.417	262383.401
total_number_of_contractions_over26.7				< 0.001
Count	71	98	169
mean	136.352	90.847	109.964
SD	94.432	68.169	83.103
Median	115.000	72.000	89.000
Q1,Q3	59.000, 187.500	41.000, 127.500	47.000, 151.000
Min	8.000	12.000	8.000
Max	441.000	338.000	441.000
total_contraction_rate_0				0.493
Count	71	98	169
mean	0.237	0.247	0.243
SD	0.168	0.153	0.159
Median	0.216	0.222	0.218
Q1,Q3	0.129, 0.293	0.122, 0.341	0.129, 0.326
Min	0.012	0.000	0.000
Max	0.970	0.685	0.970
total_contraction_rate_1_3				0.274
Count	71	98	169
mean	0.327	0.344	0.337
SD	0.087	0.093	0.090
Median	0.336	0.357	0.343
Q1,Q3	0.268, 0.389	0.272, 0.411	0.272, 0.401
Min	0.029	0.110	0.029
Max	0.488	0.566	0.566
total_contraction_rate_4_6				0.866
Count	71	98	169
mean	0.188	0.187	0.188
SD	0.065	0.071	0.068
Median	0.189	0.192	0.189
Q1,Q3	0.137, 0.227	0.133, 0.225	0.133, 0.226
Min	0.000	0.041	0.000
Max	0.366	0.364	0.366
total_contraction_rate_7_9				0.836
Count	71	98	169
mean	0.130	0.136	0.133
SD	0.071	0.082	0.077
Median	0.112	0.115	0.114
Q1,Q3	0.085, 0.165	0.081, 0.192	0.082, 0.186
Min	0.000	0.005	0.000
Max	0.336	0.455	0.455
total_contraction_rate_over10				0.010
Count	71	98	169
mean	0.119	0.085	0.099
SD	0.091	0.071	0.081
Median	0.106	0.067	0.078
Q1,Q3	0.048, 0.160	0.030, 0.125	0.038, 0.143
Min	0.000	0.000	0.000
Max	0.405	0.283	0.405
number_of_contractions_over26.7_per_hour				< 0.001
Count	71	98	169
mean	46.195	27.688	35.463
SD	35.738	23.529	30.590
Median	38.169	20.876	25.165
Q1,Q3	20.018, 59.233	11.757, 37.625	15.186, 47.064
Min	4.460	3.119	3.119
Max	168.976	155.521	168.976
imputed_total_number_of_contractions				< 0.001
Count	71	98	169
mean	1561.901	842.918	1144.976
SD	1075.089	398.189	836.553
Median	1233.000	788.000	963.000
Q1,Q3	955.000, 1826.000	577.000, 1006.000	673.000, 1336.000
Min	332.000	204.000	204.000
Max	6210.000	2526.000	6210.000
imputed_sum_of_amplitude				< 0.001
Count	71	98	169
mean	187578.734	108176.988	141535.118
SD	139333.716	38328.935	102383.712
Median	158548.129	103445.589	119039.990
Q1,Q3	115931.473, 200990.440	79719.797, 129270.590	89727.664, 160351.580
Min	62203.157	34829.877	34829.877
Max	950697.962	225832.269	950697.962
number_of_fast_epochs				0.361
Count	71	98	169
mean	4.113	4.724	4.467
SD	3.871	4.497	4.244
Median	3.000	4.000	4.000
Q1,Q3	1.000, 6.000	1.000, 7.000	1.000, 7.000
Min	0.000	0.000	0.000
Max	17.000	27.000	27.000
number_of_medium_epochs				0.013
Count	71	98	169
mean	2.592	3.531	3.136
SD	2.950	2.894	2.946
Median	2.000	3.000	2.000
Q1,Q3	1.000, 3.000	1.000, 5.000	1.000, 5.000
Min	0.000	0.000	0.000
Max	18.000	13.000	18.000
number_of_slow_epochs				0.006
Count	71	98	169
mean	3.437	4.663	4.148
SD	3.379	3.386	3.427
Median	3.000	4.000	3.000
Q1,Q3	1.000, 5.000	2.000, 6.000	1.000, 6.000
Min	0.000	0.000	0.000
Max	15.000	15.000	15.000
number_of_total_epochs				0.011
Count	71	98	169
mean	10.141	12.918	11.751
SD	7.769	8.196	8.113
Median	9.000	12.000	10.000
Q1,Q3	5.000, 14.000	7.000, 18.000	6.000, 17.000
Min	1.000	1.000	1.000
Max	37.000	51.000	51.000
number_of_fast_epochs_frequency				0.651
Count	71	98	169
mean	1.275	1.377	1.334
SD	1.138	1.188	1.165
Median	1.115	1.108	1.115
Q1,Q3	0.395, 1.765	0.378, 2.274	0.384, 1.931
Min	0.000	0.000	0.000
Max	5.168	4.601	5.168
number_of_medium_epochs_frequency				0.027
Count	71	98	169
mean	0.729	1.019	0.897
SD	0.654	0.837	0.777
Median	0.622	0.847	0.743
Q1,Q3	0.300, 0.994	0.339, 1.529	0.315, 1.268
Min	0.000	0.000	0.000
Max	3.240	4.302	4.302
number_of_slow_epochs_frequency				0.002
Count	71	98	169
mean	0.899	1.295	1.129
SD	0.722	0.850	0.820
Median	0.779	1.214	0.980
Q1,Q3	0.366, 1.406	0.689, 1.832	0.497, 1.649
Min	0.000	0.000	0.000
Max	3.232	3.648	3.648
number_of_total_epochs_frequency				0.008
Count	71	98	169
mean	2.903	3.691	3.360
SD	1.803	2.027	1.970
Median	2.512	3.430	3.091
Q1,Q3	1.689, 3.855	2.159, 4.985	1.779, 4.427
Min	0.326	0.302	0.302
Max	7.846	9.386	9.386
motility_index				0.023
Count	71	98	169
mean	18.329	18.001	18.139
SD	0.928	0.873	0.908
Median	18.397	18.086	18.147
Q1,Q3	17.833, 19.057	17.521, 18.615	17.589, 18.737
Min	15.981	15.750	15.750
Max	20.278	20.154	20.278

Control_Patients_5min_RMSSD <- Control_Patients_5min %>%
  group_by(combined_ID, epoch_type) %>%
  summarize(
    mean_RMSSD_Time = mean(epoch_RMSSD_Time),
    median_RMSSD_Time = median(epoch_RMSSD_Time),
    sd_RMSSD_Time = sd(epoch_RMSSD_Time),
    mean_RMSSD_Magnitude = mean(epoch_RMSSD_Magnitude),
    median_RMSSD_Magnitude = median(epoch_RMSSD_Magnitude),
    sd_RMSSD_Magnitude = sd(epoch_RMSSD_Magnitude),
    n_epochs = n(),
    .groups = "drop"
  ) %>%
  pivot_wider(
    id_cols = combined_ID,
    names_from = epoch_type,
    values_from = c(mean_RMSSD_Time, median_RMSSD_Time, sd_RMSSD_Time, mean_RMSSD_Magnitude, median_RMSSD_Magnitude, sd_RMSSD_Magnitude, n_epochs)) %>%
  mutate(source = "Jack's Data")

Control_Patients_5min_total_epoch <- Control_Patients_5min %>%
  group_by(combined_ID) %>%
  summarize(
    median_RMSSD_Time_total_epoch = median(epoch_RMSSD_Time),
    median_RMSSD_Magnitude_total_epoch = median(epoch_RMSSD_Magnitude),
    .groups = "drop"
  )

Control_Patients_5min_RMSSD <- Control_Patients_5min_RMSSD %>%
  left_join(Control_Patients_5min_total_epoch)

EVA_Patients_5min_RMSSD <- EVA_Patients_5min %>%
  group_by(combined_ID, epoch_type) %>%
  summarize(
    mean_RMSSD_Time = mean(epoch_RMSSD_Time),
    median_RMSSD_Time = median(epoch_RMSSD_Time),
    sd_RMSSD_Time = sd(epoch_RMSSD_Time),
    mean_RMSSD_Magnitude = mean(epoch_RMSSD_Magnitude),
    median_RMSSD_Magnitude = median(epoch_RMSSD_Magnitude),
    sd_RMSSD_Magnitude = sd(epoch_RMSSD_Magnitude),
    n_epochs = n(),
    .groups = "drop"
  ) %>%
  pivot_wider(
    id_cols = combined_ID,
    names_from = epoch_type,
    values_from = c(mean_RMSSD_Time, median_RMSSD_Time, sd_RMSSD_Time, mean_RMSSD_Magnitude, median_RMSSD_Magnitude, sd_RMSSD_Magnitude, n_epochs)
  )%>%
  mutate(source = "EVA Data")

EVA_Patients_5min_total_epoch <- EVA_Patients_5min %>%
  group_by(combined_ID) %>%
  summarize(
    median_RMSSD_Time_total_epoch = median(epoch_RMSSD_Time),
    median_RMSSD_Magnitude_total_epoch = median(epoch_RMSSD_Magnitude),
    .groups = "drop"
  )

EVA_Patients_5min_RMSSD <- EVA_Patients_5min_RMSSD %>%
  left_join(EVA_Patients_5min_total_epoch)

EVA_Controls_5min_RMSSD <- EVA_Controls_5min %>%
  group_by(combined_ID, epoch_type) %>%
  summarize(
    mean_RMSSD_Time = mean(epoch_RMSSD_Time),
    median_RMSSD_Time = median(epoch_RMSSD_Time),
    sd_RMSSD_Time = sd(epoch_RMSSD_Time),
    mean_RMSSD_Magnitude = mean(epoch_RMSSD_Magnitude),
    median_RMSSD_Magnitude = median(epoch_RMSSD_Magnitude),
    sd_RMSSD_Magnitude = sd(epoch_RMSSD_Magnitude),
    n_epochs = n(),
    .groups = "drop"
  ) %>%
  pivot_wider(
    id_cols = combined_ID,
    names_from = epoch_type,
    values_from = c(mean_RMSSD_Time, median_RMSSD_Time, sd_RMSSD_Time, mean_RMSSD_Magnitude, median_RMSSD_Magnitude, sd_RMSSD_Magnitude, n_epochs)
  ) %>%
  mutate(source = "EVA Controls")

EVA_Controls_5min_total_epoch <- EVA_Controls_5min %>%
  group_by(combined_ID) %>%
  summarize(
    median_RMSSD_Time_total_epoch = median(epoch_RMSSD_Time),
    median_RMSSD_Magnitude_total_epoch = median(epoch_RMSSD_Magnitude),
    .groups = "drop"
  )

EVA_Controls_5min_RMSSD <- EVA_Controls_5min_RMSSD %>%
  left_join(EVA_Controls_5min_total_epoch)

all_data_RMSSD <- rbind(Control_Patients_5min_RMSSD, EVA_Patients_5min_RMSSD, EVA_Controls_5min_RMSSD)

mycontrols <- tableby.control(test = T, total = TRUE,
                              numeric.test = "kwt", cat.test = "chisq",
                              numeric.stats=c("N", "Nmiss", "mean", "sd", "median", "q1q3", "min", "max"),
                              cat.stats=c("countpct"),
                              stats.labels=list(N='Count', Nmiss='Missing Values', mean='mean', sd='SD', median='Median', q1q3='Q1,Q3', min='Min', max='Max'))


tab.test <- tableby(source ~ `median_RMSSD_Time_fast epoch` + `median_RMSSD_Time_medium epoch` + `median_RMSSD_Time_slow epoch`+ median_RMSSD_Time_total_epoch + `median_RMSSD_Magnitude_fast epoch` + `median_RMSSD_Magnitude_medium epoch` + `median_RMSSD_Magnitude_slow epoch` + median_RMSSD_Magnitude_total_epoch + `n_epochs_fast epoch` + `n_epochs_medium epoch` + `n_epochs_slow epoch`, data=all_data_RMSSD, control = mycontrols)

summary(tab.test)

	EVA Controls (N=12)	EVA Data (N=71)	Jack’s Data (N=98)	Total (N=181)	p value
median_RMSSD_Time_fast epoch					0.032
Count	11	61	83	155
Missing Values	1	10	15	26
mean	52.868	53.136	50.465	51.687
SD	6.292	6.481	4.285	5.518
Median	53.760	52.540	49.280	50.290
Q1,Q3	48.420, 56.660	48.160, 57.180	47.425, 53.165	47.672, 55.375
Min	42.750	41.540	43.860	41.540
Max	64.695	71.790	63.850	71.790
median_RMSSD_Time_medium epoch					0.016
Count	8	58	87	153
Missing Values	4	13	11	28
mean	61.939	67.132	63.418	64.748
SD	7.267	9.843	7.848	8.787
Median	63.678	65.590	61.950	63.260
Q1,Q3	57.229, 66.366	60.888, 70.885	58.100, 65.892	58.990, 68.070
Min	49.140	50.480	52.110	49.140
Max	71.835	101.345	96.330	101.345
median_RMSSD_Time_slow epoch					0.117
Count	10	59	90	159
Missing Values	2	12	8	22
mean	83.365	83.876	78.458	80.777
SD	15.330	14.217	9.098	11.886
Median	80.938	80.715	77.975	79.930
Q1,Q3	73.425, 86.995	73.755, 89.062	71.345, 84.540	72.765, 86.440
Min	65.970	64.010	52.200	52.200
Max	116.840	138.230	101.360	138.230
median_RMSSD_Time_total_epoch					0.992
Count	12	71	98	181
mean	63.672	64.233	64.060	64.102
SD	9.922	12.860	10.438	11.361
Median	64.442	63.615	62.395	63.260
Q1,Q3	55.464, 69.134	57.370, 69.165	57.683, 70.236	57.280, 69.530
Min	47.390	46.380	46.620	46.380
Max	78.875	138.230	98.730	138.230
median_RMSSD_Magnitude_fast epoch					0.168
Count	11	61	83	155
Missing Values	1	10	15	26
mean	84.860	73.035	65.556	69.869
SD	48.989	28.061	24.956	28.711
Median	69.730	72.370	61.095	66.850
Q1,Q3	50.300, 107.558	54.360, 86.700	47.792, 75.675	49.400, 82.828
Min	36.570	24.450	20.160	20.160
Max	189.670	158.900	157.550	189.670
median_RMSSD_Magnitude_medium epoch					0.517
Count	8	58	87	153
Missing Values	4	13	11	28
mean	55.849	60.919	56.492	58.137
SD	21.074	25.692	23.552	24.220
Median	53.620	53.445	49.800	51.210
Q1,Q3	38.620, 72.134	42.785, 71.294	40.300, 63.960	40.820, 67.865
Min	28.360	27.860	28.345	27.860
Max	83.405	151.905	149.150	151.905
median_RMSSD_Magnitude_slow epoch					0.084
Count	10	59	90	159
Missing Values	2	12	8	22
mean	49.098	47.339	43.154	45.081
SD	24.307	24.965	26.142	25.543
Median	41.740	43.950	35.885	40.860
Q1,Q3	35.720, 53.392	34.177, 49.892	26.372, 49.825	28.727, 50.502
Min	28.680	17.460	20.100	17.460
Max	112.800	177.435	180.510	180.510
median_RMSSD_Magnitude_total_epoch					0.016
Count	12	71	98	181
mean	68.902	59.947	52.018	56.248
SD	41.847	24.122	21.697	24.755
Median	60.030	55.440	47.945	51.210
Q1,Q3	45.650, 69.062	45.390, 71.448	37.714, 59.303	39.990, 65.010
Min	33.395	22.125	21.700	21.700
Max	189.670	139.395	136.030	189.670
n_epochs_fast epoch					0.131
Count	11	61	83	155
Missing Values	1	10	15	26
mean	3.545	4.787	5.578	5.123
SD	2.979	3.769	4.370	4.076
Median	2.000	4.000	5.000	4.000
Q1,Q3	1.000, 5.000	2.000, 7.000	2.500, 7.000	2.000, 7.000
Min	1.000	1.000	1.000	1.000
Max	10.000	17.000	27.000	27.000
n_epochs_medium epoch					0.070
Count	8	58	87	153
Missing Values	4	13	11	28
mean	2.750	3.172	3.977	3.608
SD	2.053	2.968	2.766	2.831
Median	2.000	2.500	3.000	3.000
Q1,Q3	1.750, 3.000	1.000, 4.000	2.000, 6.000	2.000, 5.000
Min	1.000	1.000	1.000	1.000
Max	6.000	18.000	13.000	18.000
n_epochs_slow epoch					0.088
Count	10	59	90	159
Missing Values	2	12	8	22
mean	4.300	4.136	5.078	4.679
SD	2.312	3.293	3.219	3.215
Median	3.000	3.000	4.000	4.000
Q1,Q3	3.000, 6.500	1.500, 5.500	3.000, 7.000	2.000, 6.500
Min	1.000	1.000	1.000	1.000
Max	8.000	15.000	15.000	15.000

4.4 Demographic Table

Plot_conditioning(all_data, c("Age" , "Sex" , "MCASS_total" , "Autonomic_Neuropathy" , "Prolonged_SBTT"), "source")

mycontrols <- tableby.control(test = T, total = TRUE,
                              numeric.test = "kwt", cat.test = "chisq",
                              numeric.stats=c("N", "Nmiss", "mean", "sd", "median", "q1q3", "min", "max"),
                              cat.stats=c("countpct"),
                              stats.labels=list(N='Count', Nmiss='Missing Values', mean='mean', sd='SD', median='Median', q1q3='Q1,Q3', min='Min', max='Max'))


tab.test <- tableby(source ~ Age + Sex + MCASS_total + Autonomic_Neuropathy + Prolonged_SBTT, data=all_data, control = mycontrols)

summary(tab.test)

	EVA Patient (N=71)	Jack’s Data (N=98)	Total (N=169)	p value
Age				0.004
Count	71	97	168
Missing Values	0	1	1
mean	52.986	45.577	48.708
SD	11.819	18.271	16.246
Median	55.000	41.000	48.500
Q1,Q3	44.000, 62.500	30.000, 65.000	36.500, 64.000
Min	28.000	18.000	18.000
Max	76.000	80.000	80.000
Sex				0.002
Female	18 (25.4%)	48 (49.0%)	66 (39.1%)
Male	53 (74.6%)	50 (51.0%)	103 (60.9%)
MCASS_total
Mild AN	20 (28.2%)	0	20 (28.2%)
Moderate AN	3 (4.2%)	0	3 (4.2%)
Normal/Equivocal	48 (67.6%)	0	48 (67.6%)
Autonomic_Neuropathy
No	40 (56.3%)	0	40 (56.3%)
Yes	31 (43.7%)	0	31 (43.7%)
Prolonged_SBTT				< 0.001
No	54 (77.1%)	94 (95.9%)	148 (88.1%)
Yes	16 (22.9%)	4 (4.1%)	20 (11.9%)

mycontrols <- tableby.control(test = T, total = TRUE,
                              numeric.test = "kwt", cat.test = "chisq",
                              numeric.stats=c("N", "Nmiss", "mean", "sd", "median", "q1q3", "min", "max"),
                              cat.stats=c("countpct"),
                              stats.labels=list(N='Count', Nmiss='Missing Values', mean='mean', sd='SD', median='Median', q1q3='Q1,Q3', min='Min', max='Max'))


tab.test <- tableby(groups ~ Age + Sex + MCASS_total + Autonomic_Neuropathy + Prolonged_SBTT, data=all_data, control = mycontrols)

summary(tab.test)

	Jack’s Data (N=98)	normal_SBTT_abnormal_CASS (N=24)	normal_SBTT_normal_CASS (N=30)	prolonged_SBTT (N=16)	Total (N=168)	p value
Age						0.028
Count	97	24	30	16	167
Missing Values	1	0	0	0	1
mean	45.577	54.083	53.600	51.750	48.832
SD	18.271	12.392	10.820	12.008	16.215
Median	41.000	57.000	53.000	47.000	49.000
Q1,Q3	30.000, 65.000	48.250, 64.000	46.250, 60.000	40.000, 63.000	37.000, 64.000
Min	18.000	29.000	30.000	39.000	18.000
Max	80.000	70.000	76.000	69.000	80.000
Sex						0.012
Female	48 (49.0%)	7 (29.2%)	7 (23.3%)	3 (18.8%)	65 (38.7%)
Male	50 (51.0%)	17 (70.8%)	23 (76.7%)	13 (81.2%)	103 (61.3%)
MCASS_total
Mild AN	0	17 (70.8%)	0 (0.0%)	3 (18.8%)	20 (28.6%)
Moderate AN	0	2 (8.3%)	0 (0.0%)	1 (6.2%)	3 (4.3%)
Normal/Equivocal	0	5 (20.8%)	30 (100.0%)	12 (75.0%)	47 (67.1%)
Autonomic_Neuropathy
No	0	0 (0.0%)	30 (100.0%)	9 (56.2%)	39 (55.7%)
Yes	0	24 (100.0%)	0 (0.0%)	7 (43.8%)	31 (44.3%)
Prolonged_SBTT						< 0.001
No	94 (95.9%)	24 (100.0%)	30 (100.0%)	0 (0.0%)	148 (88.1%)
Yes	4 (4.1%)	0 (0.0%)	0 (0.0%)	16 (100.0%)	20 (11.9%)

all_data %>% filter(groups == "prolonged_SBTT") %>% pull(combined_ID) 

##  [1] 7022 7027 7037 7042 7045 7053 7055 7057 7060 7071 7114 7121 7131 7144 7146
## [16] 7161

final_spreadsheet %>% filter(ID %in% (all_data %>% filter(groups == "prolonged_SBTT") %>% pull(combined_ID))) %>% filter(as.numeric(CTT...35) > 3540) %>% select(CTT...35)

## # A tibble: 6 × 1
##   CTT...35          
##   <chr>             
## 1 5493              
## 2 3702              
## 3 5285              
## 4 5201              
## 5 21729.599999999999
## 6 5147

5 Correlation Matrix

5.1 EVA Patients

# EVA_Patients_5min_RMSSD_total <- EVA_Patients_5min %>% 
#   group_by(combined_ID) %>%
#   summarize(
#     mean_RMSSD_Time_total = mean(epoch_RMSSD_Time),
#     median_RMSSD_Time_total = median(epoch_RMSSD_Time),
#     sd_RMSSD_Time_total = sd(epoch_RMSSD_Time),
#     mean_RMSSD_Magnitude_total = mean(epoch_RMSSD_Magnitude),
#     median_RMSSD_Magnitude_total = median(epoch_RMSSD_Magnitude),
#     sd_RMSSD_Magnitude_total = sd(epoch_RMSSD_Magnitude),
#     .groups = "drop"
#   ) 

library(ggcorrplot)

# res <- cor(data.frame(
#   mean_RMSSD_Time_total = EVA_Patients_5min_RMSSD_total$mean_RMSSD_Time_total, 
#   median_RMSSD_Time_total = EVA_Patients_5min_RMSSD_total$median_RMSSD_Time_total, 
#   mean_RMSSD_Magnitude_total = EVA_Patients_5min_RMSSD_total$mean_RMSSD_Magnitude_total, 
#   median_RMSSD_Magnitude_total = EVA_Patients_5min_RMSSD_total$median_RMSSD_Magnitude_total, 
#   `cass_baseline VM SBP` = EVA_Patients_5min_RMSSD_total$`cass_baseline VM SBP`, 
#   `cass_baseline VM DBP` = EVA_Patients_5min_RMSSD_total$`cass_baseline VM DBP`, 
#   `cass_Valsalva Ratio` = EVA_Patients_5min_RMSSD_total$`cass_Valsalva Ratio`, 
#   `cass_HRDB Ratio` = EVA_Patients_5min_RMSSD_total$`cass_HRDB Ratio`, 
#   `cass_adrenergic_index` = EVA_Patients_5min_RMSSD_total$`cass_adrenergic_index`, 
#   `cass_sudomotor_index` = EVA_Patients_5min_RMSSD_total$`cass_sudomotor_index`, 
#   `cass_cardiovagal_index` = EVA_Patients_5min_RMSSD_total$`cass_cardiovagal_index`, 
#   `cass_total` = EVA_Patients_5min_RMSSD_total$`cass_total`, 
#   `mcass_brsa` = EVA_Patients_5min_RMSSD_total$`mcass_brsa`, 
#   `mcass_brsv` = EVA_Patients_5min_RMSSD_total$`mcass_brsv`, 
#   `mcass_adrenergic_index` = EVA_Patients_5min_RMSSD_total$`mcass_adrenergic_index`, 
#   `mcass_sudomotor_index` = EVA_Patients_5min_RMSSD_total$`mcass_sudomotor_index`, 
#   `mcass_cardiovagal_index` = EVA_Patients_5min_RMSSD_total$`mcass_cardiovagal_index`, 
#   `mcass_total` = EVA_Patients_5min_RMSSD_total$`mcass_total`
#   ), 
#   use = "complete.obs", method = "spearman")
# 
# # Select variables for x and y axes
# y_vars <- c("median_RMSSD_Time_total", "median_RMSSD_Magnitude_total")
# x_vars <- c("cass_baseline.VM.SBP", "cass_baseline.VM.DBP", "cass_Valsalva.Ratio", "cass_HRDB.Ratio", "cass_adrenergic_index",  "cass_sudomotor_index", "cass_cardiovagal_index", "cass_total", "mcass_brsa", "mcass_brsv", "mcass_adrenergic_index", "mcass_sudomotor_index", "mcass_cardiovagal_index", "mcass_total") 
# 
# # Subset the correlation matrix
# sub_cor <- res[y_vars, x_vars]
# 
# # Convert to long format and plot with ggplot2
library(reshape2)
# df_long <- melt(sub_cor)
# 
# ggplot(df_long, aes(Var2, Var1, fill = value)) +
#   geom_tile(color = "white") +
#   geom_text(aes(label = round(value, 2)), size = 3) +
#   scale_fill_gradient2(
#     low = "red", mid = "white", high = "blue",
#     midpoint = 0, limit = c(-1, 1), name = "Correlation"
#   ) +
#   theme_minimal() +
#   theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust = 1)) +
#   coord_fixed()

EVA_Patients_5min_RMSSD_joined <- EVA_Patients_5min_RMSSD %>% 
  left_join(final_spreadsheet, by = c("combined_ID" = "ID"))

res <- cor(data.frame(
  `median_RMSSD_Time_fast epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_fast epoch`, 
  `median_RMSSD_Time_medium epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_medium epoch`, 
  `median_RMSSD_Time_slow epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_slow epoch`, 
  `median_RMSSD_Magnitude_fast epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_fast epoch`, 
  `median_RMSSD_Magnitude_medium epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_medium epoch`, 
  `median_RMSSD_Magnitude_slow epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_slow epoch`, 
  `median_RMSSD_Time_total_epoch` = EVA_Patients_5min_RMSSD_joined$median_RMSSD_Time_total_epoch,
  `median_RMSSD_Magnitude_total_epoch` = EVA_Patients_5min_RMSSD_joined$median_RMSSD_Magnitude_total_epoch,
  `cass_baseline VM SBP` = EVA_Patients_5min_RMSSD_joined$`cass_baseline VM SBP`, 
  `cass_baseline VM DBP` = EVA_Patients_5min_RMSSD_joined$`cass_baseline VM DBP`, 
  `cass_Valsalva Ratio` = EVA_Patients_5min_RMSSD_joined$`cass_Valsalva Ratio`, 
  `cass_HRDB Ratio` = EVA_Patients_5min_RMSSD_joined$`cass_HRDB Ratio`, 
  `cass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined$`cass_adrenergic_index`, 
  `cass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined$`cass_sudomotor_index`, 
  `cass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined$`cass_cardiovagal_index`, 
  `cass_total` = EVA_Patients_5min_RMSSD_joined$`cass_total`, 
  `mcass_brsa` = EVA_Patients_5min_RMSSD_joined$`mcass_brsa`, 
  `mcass_brsv` = EVA_Patients_5min_RMSSD_joined$`mcass_brsv`, 
  `mcass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined$`mcass_adrenergic_index`, 
  `mcass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined$`mcass_sudomotor_index`, 
  `mcass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined$`mcass_cardiovagal_index`, 
  `mcass_total` = EVA_Patients_5min_RMSSD_joined$`mcass_total`
  ), 
  use = "pairwise.complete.obs", method = "spearman")

# Select variables for x and y axes
y_vars <- c("median_RMSSD_Magnitude_total_epoch", "median_RMSSD_Magnitude_slow.epoch", "median_RMSSD_Magnitude_medium.epoch", "median_RMSSD_Magnitude_fast.epoch", "median_RMSSD_Time_total_epoch", "median_RMSSD_Time_slow.epoch", "median_RMSSD_Time_medium.epoch", "median_RMSSD_Time_fast.epoch")
x_vars <- c("cass_baseline.VM.SBP", "cass_baseline.VM.DBP", "cass_Valsalva.Ratio", "cass_HRDB.Ratio", "cass_adrenergic_index",  "cass_sudomotor_index", "cass_cardiovagal_index", "cass_total", "mcass_brsa", "mcass_brsv", "mcass_adrenergic_index", "mcass_sudomotor_index", "mcass_cardiovagal_index", "mcass_total") 

# Subset the correlation matrix
sub_cor <- res[y_vars, x_vars]

# Convert to long format and plot with ggplot2
df_long <- melt(sub_cor)

ggplot(df_long, aes(Var2, Var1, fill = value)) +
  geom_tile(color = "white") +
  geom_text(aes(label = round(value, 2)), size = 3) +
  scale_fill_gradient2(
    low = "red", mid = "white", high = "blue",
    midpoint = 0, limit = c(-1, 1), name = "Correlation"
  ) +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust = 1)) +
  coord_fixed()

5.2 Prolonged SBTT

# EVA_Patients_5min_RMSSD_prolonged_SBTT <- EVA_Patients_5min_RMSSD_total %>%
#   filter(groups == "prolonged_SBTT")


# res <- cor(data.frame(
#   mean_RMSSD_Time_total = EVA_Patients_5min_RMSSD_prolonged_SBTT$mean_RMSSD_Time_total, 
#   median_RMSSD_Time_total = EVA_Patients_5min_RMSSD_prolonged_SBTT$median_RMSSD_Time_total, 
#   mean_RMSSD_Magnitude_total = EVA_Patients_5min_RMSSD_prolonged_SBTT$mean_RMSSD_Magnitude_total, 
#   median_RMSSD_Magnitude_total = EVA_Patients_5min_RMSSD_prolonged_SBTT$median_RMSSD_Magnitude_total, 
#   `cass_baseline VM SBP` = EVA_Patients_5min_RMSSD_prolonged_SBTT$`cass_baseline VM SBP`, 
#   `cass_baseline VM DBP` = EVA_Patients_5min_RMSSD_prolonged_SBTT$`cass_baseline VM DBP`, 
#   `cass_Valsalva Ratio` = EVA_Patients_5min_RMSSD_prolonged_SBTT$`cass_Valsalva Ratio`, 
#   `cass_HRDB Ratio` = EVA_Patients_5min_RMSSD_prolonged_SBTT$`cass_HRDB Ratio`, 
#   `cass_adrenergic_index` = EVA_Patients_5min_RMSSD_prolonged_SBTT$`cass_adrenergic_index`, 
#   `cass_sudomotor_index` = EVA_Patients_5min_RMSSD_prolonged_SBTT$`cass_sudomotor_index`, 
#   `cass_cardiovagal_index` = EVA_Patients_5min_RMSSD_prolonged_SBTT$`cass_cardiovagal_index`, 
#   `cass_total` = EVA_Patients_5min_RMSSD_prolonged_SBTT$`cass_total`, 
#   `mcass_brsa` = EVA_Patients_5min_RMSSD_prolonged_SBTT$`mcass_brsa`, 
#   `mcass_brsv` = EVA_Patients_5min_RMSSD_prolonged_SBTT$`mcass_brsv`, 
#   `mcass_adrenergic_index` = EVA_Patients_5min_RMSSD_prolonged_SBTT$`mcass_adrenergic_index`, 
#   `mcass_sudomotor_index` = EVA_Patients_5min_RMSSD_prolonged_SBTT$`mcass_sudomotor_index`, 
#   `mcass_cardiovagal_index` = EVA_Patients_5min_RMSSD_prolonged_SBTT$`mcass_cardiovagal_index`, 
#   `mcass_total` = EVA_Patients_5min_RMSSD_prolonged_SBTT$`mcass_total`
#   ), 
#   use = "complete.obs", method = "spearman")
# 
# # Select variables for x and y axes
# y_vars <- c( "median_RMSSD_Time_total", "median_RMSSD_Magnitude_total")
# x_vars <- c("cass_baseline.VM.SBP", "cass_baseline.VM.DBP", "cass_Valsalva.Ratio", "cass_HRDB.Ratio", "cass_adrenergic_index",  "cass_sudomotor_index", "cass_cardiovagal_index", "cass_total", "mcass_brsa", "mcass_brsv", "mcass_adrenergic_index", "mcass_sudomotor_index", "mcass_cardiovagal_index", "mcass_total") 
# 
# # Subset the correlation matrix
# sub_cor <- res[y_vars, x_vars]
# 
# # Convert to long format and plot with ggplot2
# df_long <- melt(sub_cor)
# 
# ggplot(df_long, aes(Var2, Var1, fill = value)) +
#   geom_tile(color = "white") +
#   geom_text(aes(label = round(value, 2)), size = 3) +
#   scale_fill_gradient2(
#     low = "red", mid = "white", high = "blue",
#     midpoint = 0, limit = c(-1, 1), name = "Correlation"
#   ) +
#   theme_minimal() +
#   theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust = 1)) +
#   coord_fixed()

EVA_Patients_5min_RMSSD_joined_prolonged_SBTT <- EVA_Patients_5min_RMSSD_joined %>% filter(groups == "prolonged_SBTT")

res <- cor(data.frame(
  `mean_RMSSD_Time_fast epoch` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`mean_RMSSD_Time_fast epoch`, 
  `mean_RMSSD_Time_medium epoch` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`mean_RMSSD_Time_medium epoch`, 
  `mean_RMSSD_Time_slow epoch` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`mean_RMSSD_Time_slow epoch`, 
  `median_RMSSD_Time_fast epoch` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`median_RMSSD_Time_fast epoch`, 
  `median_RMSSD_Time_medium epoch` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`median_RMSSD_Time_medium epoch`, 
  `median_RMSSD_Time_slow epoch` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`median_RMSSD_Time_slow epoch`, 
  `mean_RMSSD_Magnitude_fast epoch` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`mean_RMSSD_Magnitude_fast epoch`, 
  `mean_RMSSD_Magnitude_medium epoch` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`mean_RMSSD_Magnitude_medium epoch`, 
  `mean_RMSSD_Magnitude_slow epoch` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`mean_RMSSD_Magnitude_slow epoch`, 
  `median_RMSSD_Magnitude_fast epoch` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`median_RMSSD_Magnitude_fast epoch`, 
  `median_RMSSD_Magnitude_medium epoch` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`median_RMSSD_Magnitude_medium epoch`, 
  `median_RMSSD_Magnitude_slow epoch` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`median_RMSSD_Magnitude_slow epoch`, 
  `median_RMSSD_Time_total_epoch` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$median_RMSSD_Time_total_epoch,
  `median_RMSSD_Magnitude_total_epoch` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$median_RMSSD_Magnitude_total_epoch,
  `cass_baseline VM SBP` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`cass_baseline VM SBP`, 
  `cass_baseline VM DBP` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`cass_baseline VM DBP`, 
  `cass_Valsalva Ratio` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`cass_Valsalva Ratio`, 
  `cass_HRDB Ratio` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`cass_HRDB Ratio`, 
  `cass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`cass_adrenergic_index`, 
  `cass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`cass_sudomotor_index`, 
  `cass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`cass_cardiovagal_index`, 
  `cass_total` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`cass_total`, 
  `mcass_brsa` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`mcass_brsa`, 
  `mcass_brsv` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`mcass_brsv`, 
  `mcass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`mcass_adrenergic_index`, 
  `mcass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`mcass_sudomotor_index`, 
  `mcass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`mcass_cardiovagal_index`, 
  `mcass_total` = EVA_Patients_5min_RMSSD_joined_prolonged_SBTT$`mcass_total`
  ), 
  use = "pairwise.complete.obs", method = "spearman")

# Select variables for x and y axes
y_vars <- c("median_RMSSD_Magnitude_total_epoch", "median_RMSSD_Magnitude_slow.epoch", "median_RMSSD_Magnitude_medium.epoch", "median_RMSSD_Magnitude_fast.epoch", "median_RMSSD_Time_total_epoch", "median_RMSSD_Time_slow.epoch", "median_RMSSD_Time_medium.epoch", "median_RMSSD_Time_fast.epoch")
x_vars <- c("cass_baseline.VM.SBP", "cass_baseline.VM.DBP", "cass_Valsalva.Ratio", "cass_HRDB.Ratio", "cass_adrenergic_index",  "cass_sudomotor_index", "cass_cardiovagal_index", "cass_total", "mcass_brsa", "mcass_brsv", "mcass_adrenergic_index", "mcass_sudomotor_index", "mcass_cardiovagal_index", "mcass_total") 

# Subset the correlation matrix
sub_cor <- res[y_vars, x_vars]

# Convert to long format and plot with ggplot2
df_long <- melt(sub_cor)

ggplot(df_long, aes(Var2, Var1, fill = value)) +
  geom_tile(color = "white") +
  geom_text(aes(label = round(value, 2)), size = 3) +
  scale_fill_gradient2(
    low = "red", mid = "white", high = "blue",
    midpoint = 0, limit = c(-1, 1), name = "Correlation"
  ) +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust = 1)) +
  coord_fixed()

5.3 normal_SBTT_normal_CASS

# EVA_Patients_5min_RMSSD_normal_SBTT_normal_CASS <- EVA_Patients_5min_RMSSD_total %>%
#   filter(groups == "normal_SBTT_normal_CASS")


# res <- cor(data.frame(
#   mean_RMSSD_Time_total = EVA_Patients_5min_RMSSD_normal_SBTT_normal_CASS$mean_RMSSD_Time_total, 
#   median_RMSSD_Time_total = EVA_Patients_5min_RMSSD_normal_SBTT_normal_CASS$median_RMSSD_Time_total, 
#   mean_RMSSD_Magnitude_total = EVA_Patients_5min_RMSSD_normal_SBTT_normal_CASS$mean_RMSSD_Magnitude_total, 
#   median_RMSSD_Magnitude_total = EVA_Patients_5min_RMSSD_normal_SBTT_normal_CASS$median_RMSSD_Magnitude_total, 
#   `cass_baseline VM SBP` = EVA_Patients_5min_RMSSD_normal_SBTT_normal_CASS$`cass_baseline VM SBP`, 
#   `cass_baseline VM DBP` = EVA_Patients_5min_RMSSD_normal_SBTT_normal_CASS$`cass_baseline VM DBP`, 
#   `cass_Valsalva Ratio` = EVA_Patients_5min_RMSSD_normal_SBTT_normal_CASS$`cass_Valsalva Ratio`, 
#   `cass_HRDB Ratio` = EVA_Patients_5min_RMSSD_normal_SBTT_normal_CASS$`cass_HRDB Ratio`, 
#   `cass_adrenergic_index` = EVA_Patients_5min_RMSSD_normal_SBTT_normal_CASS$`cass_adrenergic_index`, 
#   `cass_sudomotor_index` = EVA_Patients_5min_RMSSD_normal_SBTT_normal_CASS$`cass_sudomotor_index`, 
#   `cass_cardiovagal_index` = EVA_Patients_5min_RMSSD_normal_SBTT_normal_CASS$`cass_cardiovagal_index`, 
#   `cass_total` = EVA_Patients_5min_RMSSD_normal_SBTT_normal_CASS$`cass_total`, 
#   `mcass_brsa` = EVA_Patients_5min_RMSSD_normal_SBTT_normal_CASS$`mcass_brsa`, 
#   `mcass_brsv` = EVA_Patients_5min_RMSSD_normal_SBTT_normal_CASS$`mcass_brsv`, 
#   `mcass_adrenergic_index` = EVA_Patients_5min_RMSSD_normal_SBTT_normal_CASS$`mcass_adrenergic_index`, 
#   `mcass_sudomotor_index` = EVA_Patients_5min_RMSSD_normal_SBTT_normal_CASS$`mcass_sudomotor_index`, 
#   `mcass_cardiovagal_index` = EVA_Patients_5min_RMSSD_normal_SBTT_normal_CASS$`mcass_cardiovagal_index`, 
#   `mcass_total` = EVA_Patients_5min_RMSSD_normal_SBTT_normal_CASS$`mcass_total`
#   ), 
#   use = "complete.obs", method = "spearman")
# 
# # Select variables for x and y axes
# y_vars <- c("median_RMSSD_Time_total", "median_RMSSD_Magnitude_total")
# x_vars <- c("cass_baseline.VM.SBP", "cass_baseline.VM.DBP", "cass_Valsalva.Ratio", "cass_HRDB.Ratio", "cass_adrenergic_index",  "cass_sudomotor_index", "cass_cardiovagal_index", "cass_total", "mcass_brsa", "mcass_brsv", "mcass_adrenergic_index", "mcass_sudomotor_index", "mcass_cardiovagal_index", "mcass_total") 
# 
# # Subset the correlation matrix
# sub_cor <- res[y_vars, x_vars]
# 
# # Convert to long format and plot with ggplot2
# library(reshape2)
# df_long <- melt(sub_cor)
# 
# ggplot(df_long, aes(Var2, Var1, fill = value)) +
#   geom_tile(color = "white") +
#   geom_text(aes(label = round(value, 2)), size = 3) +
#   scale_fill_gradient2(
#     low = "red", mid = "white", high = "blue",
#     midpoint = 0, limit = c(-1, 1), name = "Correlation"
#   ) +
#   theme_minimal() +
#   theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust = 1)) +
#   coord_fixed()

EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS <- EVA_Patients_5min_RMSSD_joined %>% filter(groups == "normal_SBTT_normal_CASS")

res <- cor(data.frame(
  `mean_RMSSD_Time_fast epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`mean_RMSSD_Time_fast epoch`, 
  `mean_RMSSD_Time_medium epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`mean_RMSSD_Time_medium epoch`, 
  `mean_RMSSD_Time_slow epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`mean_RMSSD_Time_slow epoch`, 
  `median_RMSSD_Time_fast epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`median_RMSSD_Time_fast epoch`, 
  `median_RMSSD_Time_medium epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`median_RMSSD_Time_medium epoch`, 
  `median_RMSSD_Time_slow epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`median_RMSSD_Time_slow epoch`, 
  `mean_RMSSD_Magnitude_fast epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`mean_RMSSD_Magnitude_fast epoch`, 
  `mean_RMSSD_Magnitude_medium epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`mean_RMSSD_Magnitude_medium epoch`, 
  `mean_RMSSD_Magnitude_slow epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`mean_RMSSD_Magnitude_slow epoch`, 
  `median_RMSSD_Magnitude_fast epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`median_RMSSD_Magnitude_fast epoch`, 
  `median_RMSSD_Magnitude_medium epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`median_RMSSD_Magnitude_medium epoch`, 
  `median_RMSSD_Magnitude_slow epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`median_RMSSD_Magnitude_slow epoch`, 
  `median_RMSSD_Time_total_epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$median_RMSSD_Time_total_epoch,
  `median_RMSSD_Magnitude_total_epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$median_RMSSD_Magnitude_total_epoch,
  `cass_baseline VM SBP` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`cass_baseline VM SBP`, 
  `cass_baseline VM DBP` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`cass_baseline VM DBP`, 
  `cass_Valsalva Ratio` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`cass_Valsalva Ratio`, 
  `cass_HRDB Ratio` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`cass_HRDB Ratio`, 
  `cass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`cass_adrenergic_index`, 
  `cass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`cass_sudomotor_index`, 
  `cass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`cass_cardiovagal_index`, 
  `cass_total` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`cass_total`, 
  `mcass_brsa` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`mcass_brsa`, 
  `mcass_brsv` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`mcass_brsv`, 
  `mcass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`mcass_adrenergic_index`, 
  `mcass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`mcass_sudomotor_index`, 
  `mcass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`mcass_cardiovagal_index`, 
  `mcass_total` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_normal_CASS$`mcass_total`
  ), 
  use = "pairwise.complete.obs", method = "spearman")

# Select variables for x and y axes
y_vars <- c("median_RMSSD_Magnitude_total_epoch", "median_RMSSD_Magnitude_slow.epoch", "median_RMSSD_Magnitude_medium.epoch", "median_RMSSD_Magnitude_fast.epoch", "median_RMSSD_Time_total_epoch", "median_RMSSD_Time_slow.epoch", "median_RMSSD_Time_medium.epoch", "median_RMSSD_Time_fast.epoch")
x_vars <- c("cass_baseline.VM.SBP", "cass_baseline.VM.DBP", "cass_Valsalva.Ratio", "cass_HRDB.Ratio", "cass_adrenergic_index",  "cass_sudomotor_index", "cass_cardiovagal_index", "cass_total", "mcass_brsa", "mcass_brsv", "mcass_adrenergic_index", "mcass_sudomotor_index", "mcass_cardiovagal_index", "mcass_total") 

# Subset the correlation matrix
sub_cor <- res[y_vars, x_vars]

# Convert to long format and plot with ggplot2
df_long <- melt(sub_cor)

ggplot(df_long, aes(Var2, Var1, fill = value)) +
  geom_tile(color = "white") +
  geom_text(aes(label = round(value, 2)), size = 3) +
  scale_fill_gradient2(
    low = "red", mid = "white", high = "blue",
    midpoint = 0, limit = c(-1, 1), name = "Correlation"
  ) +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust = 1)) +
  coord_fixed()

5.4 normal_SBTT_abnormal_CASS

# EVA_Patients_5min_RMSSD_normal_SBTT_abnormal_CASS <- EVA_Patients_5min_RMSSD_total %>%
#   filter(groups == "normal_SBTT_abnormal_CASS")

# res <- cor(data.frame(
#   mean_RMSSD_Time_total = EVA_Patients_5min_RMSSD_normal_SBTT_abnormal_CASS$mean_RMSSD_Time_total, 
#   median_RMSSD_Time_total = EVA_Patients_5min_RMSSD_normal_SBTT_abnormal_CASS$median_RMSSD_Time_total, 
#   mean_RMSSD_Magnitude_total = EVA_Patients_5min_RMSSD_normal_SBTT_abnormal_CASS$mean_RMSSD_Magnitude_total, 
#   median_RMSSD_Magnitude_total = EVA_Patients_5min_RMSSD_normal_SBTT_abnormal_CASS$median_RMSSD_Magnitude_total, 
#   `cass_baseline VM SBP` = EVA_Patients_5min_RMSSD_normal_SBTT_abnormal_CASS$`cass_baseline VM SBP`, 
#   `cass_baseline VM DBP` = EVA_Patients_5min_RMSSD_normal_SBTT_abnormal_CASS$`cass_baseline VM DBP`, 
#   `cass_Valsalva Ratio` = EVA_Patients_5min_RMSSD_normal_SBTT_abnormal_CASS$`cass_Valsalva Ratio`, 
#   `cass_HRDB Ratio` = EVA_Patients_5min_RMSSD_normal_SBTT_abnormal_CASS$`cass_HRDB Ratio`, 
#   `cass_adrenergic_index` = EVA_Patients_5min_RMSSD_normal_SBTT_abnormal_CASS$`cass_adrenergic_index`, 
#   `cass_sudomotor_index` = EVA_Patients_5min_RMSSD_normal_SBTT_abnormal_CASS$`cass_sudomotor_index`, 
#   `cass_cardiovagal_index` = EVA_Patients_5min_RMSSD_normal_SBTT_abnormal_CASS$`cass_cardiovagal_index`, 
#   `cass_total` = EVA_Patients_5min_RMSSD_normal_SBTT_abnormal_CASS$`cass_total`, 
#   `mcass_brsa` = EVA_Patients_5min_RMSSD_normal_SBTT_abnormal_CASS$`mcass_brsa`, 
#   `mcass_brsv` = EVA_Patients_5min_RMSSD_normal_SBTT_abnormal_CASS$`mcass_brsv`, 
#   `mcass_adrenergic_index` = EVA_Patients_5min_RMSSD_normal_SBTT_abnormal_CASS$`mcass_adrenergic_index`, 
#   `mcass_sudomotor_index` = EVA_Patients_5min_RMSSD_normal_SBTT_abnormal_CASS$`mcass_sudomotor_index`, 
#   `mcass_cardiovagal_index` = EVA_Patients_5min_RMSSD_normal_SBTT_abnormal_CASS$`mcass_cardiovagal_index`, 
#   `mcass_total` = EVA_Patients_5min_RMSSD_normal_SBTT_abnormal_CASS$`mcass_total`
#   ), 
#   use = "complete.obs", method = "spearman")
# 
# # Select variables for x and y axes
# y_vars <- c("median_RMSSD_Time_total", "median_RMSSD_Magnitude_total")
# x_vars <- c("cass_baseline.VM.SBP", "cass_baseline.VM.DBP", "cass_Valsalva.Ratio", "cass_HRDB.Ratio", "cass_adrenergic_index",  "cass_sudomotor_index", "cass_cardiovagal_index", "cass_total", "mcass_brsa", "mcass_brsv", "mcass_adrenergic_index", "mcass_sudomotor_index", "mcass_cardiovagal_index", "mcass_total") 
# 
# # Subset the correlation matrix
# sub_cor <- res[y_vars, x_vars]
# 
# # Convert to long format and plot with ggplot2
# library(reshape2)
# df_long <- melt(sub_cor)
# 
# ggplot(df_long, aes(Var2, Var1, fill = value)) +
#   geom_tile(color = "white") +
#   geom_text(aes(label = round(value, 2)), size = 3) +
#   scale_fill_gradient2(
#     low = "red", mid = "white", high = "blue",
#     midpoint = 0, limit = c(-1, 1), name = "Correlation"
#   ) +
#   theme_minimal() +
#   theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust = 1)) +
#   coord_fixed()

EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS <- EVA_Patients_5min_RMSSD_joined %>% filter(groups == "normal_SBTT_abnormal_CASS")

res <- cor(data.frame(
  `mean_RMSSD_Time_fast epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`mean_RMSSD_Time_fast epoch`, 
  `mean_RMSSD_Time_medium epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`mean_RMSSD_Time_medium epoch`, 
  `mean_RMSSD_Time_slow epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`mean_RMSSD_Time_slow epoch`, 
  `median_RMSSD_Time_fast epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`median_RMSSD_Time_fast epoch`, 
  `median_RMSSD_Time_medium epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`median_RMSSD_Time_medium epoch`, 
  `median_RMSSD_Time_slow epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`median_RMSSD_Time_slow epoch`, 
  `mean_RMSSD_Magnitude_fast epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`mean_RMSSD_Magnitude_fast epoch`, 
  `mean_RMSSD_Magnitude_medium epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`mean_RMSSD_Magnitude_medium epoch`, 
  `mean_RMSSD_Magnitude_slow epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`mean_RMSSD_Magnitude_slow epoch`, 
  `median_RMSSD_Magnitude_fast epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`median_RMSSD_Magnitude_fast epoch`, 
  `median_RMSSD_Magnitude_medium epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`median_RMSSD_Magnitude_medium epoch`, 
  `median_RMSSD_Magnitude_slow epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`median_RMSSD_Magnitude_slow epoch`,  
  `median_RMSSD_Time_total_epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$median_RMSSD_Time_total_epoch,
  `median_RMSSD_Magnitude_total_epoch` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$median_RMSSD_Magnitude_total_epoch,
  `cass_baseline VM SBP` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`cass_baseline VM SBP`, 
  `cass_baseline VM DBP` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`cass_baseline VM DBP`, 
  `cass_Valsalva Ratio` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`cass_Valsalva Ratio`, 
  `cass_HRDB Ratio` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`cass_HRDB Ratio`, 
  `cass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`cass_adrenergic_index`, 
  `cass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`cass_sudomotor_index`, 
  `cass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`cass_cardiovagal_index`, 
  `cass_total` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`cass_total`, 
  `mcass_brsa` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`mcass_brsa`, 
  `mcass_brsv` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`mcass_brsv`, 
  `mcass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`mcass_adrenergic_index`, 
  `mcass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`mcass_sudomotor_index`, 
  `mcass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`mcass_cardiovagal_index`, 
  `mcass_total` = EVA_Patients_5min_RMSSD_joined_normal_SBTT_abnormal_CASS$`mcass_total`
  ), 
  use = "pairwise.complete.obs", method = "spearman")

# Select variables for x and y axes
y_vars <- c("median_RMSSD_Magnitude_total_epoch", "median_RMSSD_Magnitude_slow.epoch", "median_RMSSD_Magnitude_medium.epoch", "median_RMSSD_Magnitude_fast.epoch", "median_RMSSD_Time_total_epoch", "median_RMSSD_Time_slow.epoch", "median_RMSSD_Time_medium.epoch", "median_RMSSD_Time_fast.epoch")
x_vars <- c("cass_baseline.VM.SBP", "cass_baseline.VM.DBP", "cass_Valsalva.Ratio", "cass_HRDB.Ratio", "cass_adrenergic_index",  "cass_sudomotor_index", "cass_cardiovagal_index", "cass_total", "mcass_brsa", "mcass_brsv", "mcass_adrenergic_index", "mcass_sudomotor_index", "mcass_cardiovagal_index", "mcass_total") 

# Subset the correlation matrix
sub_cor <- res[y_vars, x_vars]

# Convert to long format and plot with ggplot2
df_long <- melt(sub_cor)

ggplot(df_long, aes(Var2, Var1, fill = value)) +
  geom_tile(color = "white") +
  geom_text(aes(label = round(value, 2)), size = 3) +
  scale_fill_gradient2(
    low = "red", mid = "white", high = "blue",
    midpoint = 0, limit = c(-1, 1), name = "Correlation"
  ) +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust = 1)) +
  coord_fixed()

5.5 EVA Controls

# EVA_Controls_5min_RMSSD_total <- EVA_Controls_5min %>% 
#   group_by(combined_ID) %>%
#   summarize(
#     mean_RMSSD_Time_total = mean(epoch_RMSSD_Time),
#     median_RMSSD_Time_total = median(epoch_RMSSD_Time),
#     sd_RMSSD_Time_total = sd(epoch_RMSSD_Time),
#     mean_RMSSD_Magnitude_total = mean(epoch_RMSSD_Magnitude),
#     median_RMSSD_Magnitude_total = median(epoch_RMSSD_Magnitude),
#     sd_RMSSD_Magnitude_total = sd(epoch_RMSSD_Magnitude),
#     .groups = "drop"
#   ) 
# 
# EVA_Controls_5min_RMSSD_total <- EVA_Controls_5min_RMSSD_total %>% 
#   left_join(final_spreadsheet_control, by = c("combined_ID" = "SUBJID"))


# res <- cor(data.frame(
#   mean_RMSSD_Time_total = EVA_Controls_5min_RMSSD_total$mean_RMSSD_Time_total, 
#   median_RMSSD_Time_total = EVA_Controls_5min_RMSSD_total$median_RMSSD_Time_total, 
#   mean_RMSSD_Magnitude_total = EVA_Controls_5min_RMSSD_total$mean_RMSSD_Magnitude_total, 
#   median_RMSSD_Magnitude_total = EVA_Controls_5min_RMSSD_total$median_RMSSD_Magnitude_total, 
#   `cass_baseline SBP` = EVA_Controls_5min_RMSSD_total$`cass_baseline SBP`, 
#   `cass_baseline DBP` = EVA_Controls_5min_RMSSD_total$`cass_baseline DBP`, 
#   `cass_Valsalva Ratio` = EVA_Controls_5min_RMSSD_total$`cass_valsalva ratio`, 
#   `cass_HRDB Ratio` = EVA_Controls_5min_RMSSD_total$`cass_HRDB ratio`, 
#   `cass_adrenergic_index` = EVA_Controls_5min_RMSSD_total$`cass_adrenergic_index`, 
#   `cass_sudomotor_index` = EVA_Controls_5min_RMSSD_total$`cass_sudomotor_index`, 
#   `cass_cardiovagal_index` = EVA_Controls_5min_RMSSD_total$`cass_cardiovagal_index`, 
#   `cass_total` = EVA_Controls_5min_RMSSD_total$`cass_total`, 
#   `mcass_brsa` = EVA_Controls_5min_RMSSD_total$`mcass_brsa`, 
#   `mcass_brsv` = EVA_Controls_5min_RMSSD_total$`mcass_brsv`, 
#   `mcass_adrenergic_index` = EVA_Controls_5min_RMSSD_total$`mcass_adrenergic_index`, 
#   `mcass_sudomotor_index` = EVA_Controls_5min_RMSSD_total$`mcass_sudomotor_index`, 
#   `mcass_cardiovagal_index` = EVA_Controls_5min_RMSSD_total$`mcass_cardiovagal_index`, 
#   `mcass_total` = EVA_Controls_5min_RMSSD_total$`mcass_total`
#   ), 
#   use = "complete.obs", method = "spearman")
# 
# # Select variables for x and y axes
# y_vars <- c("median_RMSSD_Time_total", "median_RMSSD_Magnitude_total")
# x_vars <- c("cass_baseline.SBP", "cass_baseline.DBP", "cass_Valsalva.Ratio", "cass_HRDB.Ratio", "cass_adrenergic_index",  "cass_sudomotor_index", "cass_cardiovagal_index", "cass_total", "mcass_brsa", "mcass_brsv", "mcass_adrenergic_index", "mcass_sudomotor_index", "mcass_cardiovagal_index", "mcass_total") 
# 
# # Subset the correlation matrix
# sub_cor <- res[y_vars, x_vars]
# 
# # Convert to long format and plot with ggplot2
# df_long <- melt(sub_cor)
# 
# ggplot(df_long, aes(Var2, Var1, fill = value)) +
#   geom_tile(color = "white") +
#   geom_text(aes(label = round(value, 2)), size = 3) +
#   scale_fill_gradient2(
#     low = "red", mid = "white", high = "blue",
#     midpoint = 0, limit = c(-1, 1), name = "Correlation"
#   ) +
#   theme_minimal() +
#   theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust = 1)) +
#   coord_fixed()

EVA_Controls_5min_RMSSD_joined <- EVA_Controls_5min_RMSSD %>% 
  left_join(final_spreadsheet_control, by = c("combined_ID" = "SUBJID"))

res <- cor(data.frame(
  `mean_RMSSD_Time_fast epoch` = EVA_Controls_5min_RMSSD_joined$`mean_RMSSD_Time_fast epoch`, 
  `mean_RMSSD_Time_medium epoch` = EVA_Controls_5min_RMSSD_joined$`mean_RMSSD_Time_medium epoch`, 
  `mean_RMSSD_Time_slow epoch` = EVA_Controls_5min_RMSSD_joined$`mean_RMSSD_Time_slow epoch`, 
  `median_RMSSD_Time_fast epoch` = EVA_Controls_5min_RMSSD_joined$`median_RMSSD_Time_fast epoch`, 
  `median_RMSSD_Time_medium epoch` = EVA_Controls_5min_RMSSD_joined$`median_RMSSD_Time_medium epoch`, 
  `median_RMSSD_Time_slow epoch` = EVA_Controls_5min_RMSSD_joined$`median_RMSSD_Time_slow epoch`, 
  `mean_RMSSD_Magnitude_fast epoch` = EVA_Controls_5min_RMSSD_joined$`mean_RMSSD_Magnitude_fast epoch`, 
  `mean_RMSSD_Magnitude_medium epoch` = EVA_Controls_5min_RMSSD_joined$`mean_RMSSD_Magnitude_medium epoch`, 
  `mean_RMSSD_Magnitude_slow epoch` = EVA_Controls_5min_RMSSD_joined$`mean_RMSSD_Magnitude_slow epoch`, 
  `median_RMSSD_Magnitude_fast epoch` = EVA_Controls_5min_RMSSD_joined$`median_RMSSD_Magnitude_fast epoch`, 
  `median_RMSSD_Magnitude_medium epoch` = EVA_Controls_5min_RMSSD_joined$`median_RMSSD_Magnitude_medium epoch`, 
  `median_RMSSD_Magnitude_slow epoch` = EVA_Controls_5min_RMSSD_joined$`median_RMSSD_Magnitude_slow epoch`, 
  `median_RMSSD_Time_total_epoch` = EVA_Controls_5min_RMSSD_joined$median_RMSSD_Time_total_epoch,
  `median_RMSSD_Magnitude_total_epoch` = EVA_Controls_5min_RMSSD_joined$median_RMSSD_Magnitude_total_epoch,
  `cass_baseline SBP` = EVA_Controls_5min_RMSSD_joined$`cass_baseline SBP`, 
  `cass_baseline DBP` = EVA_Controls_5min_RMSSD_joined$`cass_baseline DBP`, 
  `cass_Valsalva Ratio` = EVA_Controls_5min_RMSSD_joined$`cass_valsalva ratio`, 
  `cass_HRDB Ratio` = EVA_Controls_5min_RMSSD_joined$`cass_HRDB ratio`, 
  `cass_adrenergic_index` = EVA_Controls_5min_RMSSD_joined$`cass_adrenergic_index`, 
  `cass_sudomotor_index` = EVA_Controls_5min_RMSSD_joined$`cass_sudomotor_index`, 
  `cass_cardiovagal_index` = EVA_Controls_5min_RMSSD_joined$`cass_cardiovagal_index`, 
  `cass_total` = EVA_Controls_5min_RMSSD_joined$`cass_total`, 
  `mcass_brsa` = EVA_Controls_5min_RMSSD_joined$`mcass_brsa`, 
  `mcass_brsv` = EVA_Controls_5min_RMSSD_joined$`mcass_brsv`, 
  `mcass_adrenergic_index` = EVA_Controls_5min_RMSSD_joined$`mcass_adrenergic_index`, 
  `mcass_sudomotor_index` = EVA_Controls_5min_RMSSD_joined$`mcass_sudomotor_index`, 
  `mcass_cardiovagal_index` = EVA_Controls_5min_RMSSD_joined$`mcass_cardiovagal_index`, 
  `mcass_total` = EVA_Controls_5min_RMSSD_joined$`mcass_total`
  ), 
  use = "pairwise.complete.obs", method = "spearman")

# Select variables for x and y axes
y_vars <- c("median_RMSSD_Magnitude_total_epoch", "median_RMSSD_Magnitude_slow.epoch", "median_RMSSD_Magnitude_medium.epoch", "median_RMSSD_Magnitude_fast.epoch", "median_RMSSD_Time_total_epoch", "median_RMSSD_Time_slow.epoch", "median_RMSSD_Time_medium.epoch", "median_RMSSD_Time_fast.epoch")
x_vars <- c("cass_baseline.SBP", "cass_baseline.DBP", "cass_Valsalva.Ratio", "cass_HRDB.Ratio", "cass_adrenergic_index",  "cass_sudomotor_index", "cass_cardiovagal_index", "cass_total", "mcass_brsa", "mcass_brsv", "mcass_adrenergic_index", "mcass_sudomotor_index", "mcass_cardiovagal_index", "mcass_total") 

# Subset the correlation matrix
sub_cor <- res[y_vars, x_vars]

# Convert to long format and plot with ggplot2
df_long <- melt(sub_cor)

ggplot(df_long, aes(Var2, Var1, fill = value)) +
  geom_tile(color = "white") +
  geom_text(aes(label = round(value, 2)), size = 3) +
  scale_fill_gradient2(
    low = "red", mid = "white", high = "blue",
    midpoint = 0, limit = c(-1, 1), name = "Correlation"
  ) +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust = 1)) +
  coord_fixed()

5.6 New Correlation Matrix 1

EVA_Patients_5min_general <- EVA_Patients_5min %>% 
  group_by(combined_ID) %>% 
  slice(1) %>% 
  select(-c(Age, Sex, SBTT...34, groups)) %>%
  left_join(final_spreadsheet, by = c("combined_ID" = "ID"))

res <- cor(data.frame(
  total_contractions_frequency = EVA_Patients_5min_general$total_contractions_frequency,
  total_contraction_rate_over10 = EVA_Patients_5min_general$total_contraction_rate_over10,
  number_of_contractions_over26.7_per_hour = EVA_Patients_5min_general$number_of_contractions_over26.7_per_hour,
  number_of_total_epochs_frequency = EVA_Patients_5min_general$number_of_total_epochs_frequency,
  imputed_sum_of_amplitude = EVA_Patients_5min_general$imputed_sum_of_amplitude,
  motility_index =EVA_Patients_5min_general$motility_index,
  `cass_baseline VM SBP` = EVA_Patients_5min_general$`cass_baseline VM SBP`, 
  `cass_baseline VM DBP` = EVA_Patients_5min_general$`cass_baseline VM DBP`, 
  `cass_Valsalva Ratio` = EVA_Patients_5min_general$`cass_Valsalva Ratio`, 
  `cass_HRDB Ratio` = EVA_Patients_5min_general$`cass_HRDB Ratio`, 
  `cass_adrenergic_index` = EVA_Patients_5min_general$`cass_adrenergic_index`, 
  `cass_sudomotor_index` = EVA_Patients_5min_general$`cass_sudomotor_index`, 
  `cass_cardiovagal_index` = EVA_Patients_5min_general$`cass_cardiovagal_index`, 
  `cass_total` = EVA_Patients_5min_general$`cass_total`, 
  `mcass_brsa` = EVA_Patients_5min_general$`mcass_brsa`, 
  `mcass_brsv` = EVA_Patients_5min_general$`mcass_brsv`, 
  `mcass_adrenergic_index` = EVA_Patients_5min_general$`mcass_adrenergic_index`, 
  `mcass_sudomotor_index` = EVA_Patients_5min_general$`mcass_sudomotor_index`, 
  `mcass_cardiovagal_index` = EVA_Patients_5min_general$`mcass_cardiovagal_index`, 
  `mcass_total` = EVA_Patients_5min_general$`mcass_total`  
  ),
  use = "pairwise.complete.obs", method = "spearman")

# Select variables for x and y axes
y_vars <- c("total_contractions_frequency", "total_contraction_rate_over10", "number_of_contractions_over26.7_per_hour", "number_of_total_epochs_frequency", "imputed_sum_of_amplitude", "motility_index")
x_vars <- c("cass_adrenergic_index",  "cass_sudomotor_index", "mcass_adrenergic_index", "mcass_sudomotor_index") 



# Subset the correlation matrix
sub_cor <- res[x_vars, y_vars]

# Convert to long format and plot with ggplot2
df_long <- melt(sub_cor)

ggplot(df_long, aes(Var2, Var1, fill = value)) +
  geom_tile(color = "white") +
  geom_text(aes(label = round(value, 2)), size = 3) +
  scale_fill_gradient2(
    low = "red", mid = "white", high = "blue",
    midpoint = 0, limit = c(-1, 1), name = "Correlation"
  ) +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust = 1)) +
  coord_fixed()

res <- cor(data.frame(
  `median_RMSSD_Time_fast epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_fast epoch`, 
  `median_RMSSD_Time_medium epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_medium epoch`, 
  `median_RMSSD_Time_slow epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_slow epoch`, 
  `median_RMSSD_Magnitude_fast epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_fast epoch`, 
  `median_RMSSD_Magnitude_medium epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_medium epoch`, 
  `median_RMSSD_Magnitude_slow epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_slow epoch`, 
  `median_RMSSD_Time_total_epoch` = EVA_Patients_5min_RMSSD_joined$median_RMSSD_Time_total_epoch,
  `median_RMSSD_Magnitude_total_epoch` = EVA_Patients_5min_RMSSD_joined$median_RMSSD_Magnitude_total_epoch,
  `cass_baseline VM SBP` = EVA_Patients_5min_RMSSD_joined$`cass_baseline VM SBP`, 
  `cass_baseline VM DBP` = EVA_Patients_5min_RMSSD_joined$`cass_baseline VM DBP`, 
  `cass_Valsalva Ratio` = EVA_Patients_5min_RMSSD_joined$`cass_Valsalva Ratio`, 
  `cass_HRDB Ratio` = EVA_Patients_5min_RMSSD_joined$`cass_HRDB Ratio`, 
  `cass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined$`cass_adrenergic_index`, 
  `cass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined$`cass_sudomotor_index`, 
  `cass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined$`cass_cardiovagal_index`, 
  `cass_total` = EVA_Patients_5min_RMSSD_joined$`cass_total`, 
  `mcass_brsa` = EVA_Patients_5min_RMSSD_joined$`mcass_brsa`, 
  `mcass_brsv` = EVA_Patients_5min_RMSSD_joined$`mcass_brsv`, 
  `mcass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined$`mcass_adrenergic_index`, 
  `mcass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined$`mcass_sudomotor_index`, 
  `mcass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined$`mcass_cardiovagal_index`, 
  `mcass_total` = EVA_Patients_5min_RMSSD_joined$`mcass_total`
  ), 
  use = "pairwise.complete.obs", method = "spearman")

# Select variables for x and y axes
y_vars <- c("median_RMSSD_Magnitude_total_epoch", "median_RMSSD_Magnitude_slow.epoch", "median_RMSSD_Magnitude_medium.epoch", "median_RMSSD_Magnitude_fast.epoch", "median_RMSSD_Time_total_epoch", "median_RMSSD_Time_slow.epoch", "median_RMSSD_Time_medium.epoch", "median_RMSSD_Time_fast.epoch")
x_vars <- c("cass_adrenergic_index",  "cass_sudomotor_index", "mcass_adrenergic_index", "mcass_sudomotor_index") 



# Subset the correlation matrix
sub_cor <- res[x_vars, y_vars]

# Convert to long format and plot with ggplot2
df_long <- melt(sub_cor)

ggplot(df_long, aes(Var2, Var1, fill = value)) +
  geom_tile(color = "white") +
  geom_text(aes(label = round(value, 2)), size = 3) +
  scale_fill_gradient2(
    low = "red", mid = "white", high = "blue",
    midpoint = 0, limit = c(-1, 1), name = "Correlation"
  ) +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust = 1)) +
  coord_fixed()

5.7 New Correlation Matrix 2

EVA_Patients_5min_general_excluding_normal <- EVA_Patients_5min_general %>%
  filter(groups != "normal_SBTT_normal_CASS")

res <- cor(data.frame(
  total_contractions_frequency = EVA_Patients_5min_general_excluding_normal$total_contractions_frequency,
  total_contraction_rate_over10 = EVA_Patients_5min_general_excluding_normal$total_contraction_rate_over10,
  number_of_contractions_over26.7_per_hour = EVA_Patients_5min_general_excluding_normal$number_of_contractions_over26.7_per_hour,
  number_of_total_epochs_frequency = EVA_Patients_5min_general_excluding_normal$number_of_total_epochs_frequency,
  imputed_sum_of_amplitude = EVA_Patients_5min_general_excluding_normal$imputed_sum_of_amplitude,
  motility_index =EVA_Patients_5min_general_excluding_normal$motility_index,
  `cass_baseline VM SBP` = EVA_Patients_5min_general_excluding_normal$`cass_baseline VM SBP`,
  `cass_baseline VM DBP` = EVA_Patients_5min_general_excluding_normal$`cass_baseline VM DBP`,
  `cass_Valsalva Ratio` = EVA_Patients_5min_general_excluding_normal$`cass_Valsalva Ratio`,
  `cass_HRDB Ratio` = EVA_Patients_5min_general_excluding_normal$`cass_HRDB Ratio`,
  `cass_adrenergic_index` = EVA_Patients_5min_general_excluding_normal$`cass_adrenergic_index`,
  `cass_sudomotor_index` = EVA_Patients_5min_general_excluding_normal$`cass_sudomotor_index`,
  `cass_cardiovagal_index` = EVA_Patients_5min_general_excluding_normal$`cass_cardiovagal_index`,
  `cass_total` = EVA_Patients_5min_general_excluding_normal$`cass_total`,
  `mcass_brsa` = EVA_Patients_5min_general_excluding_normal$`mcass_brsa`,
  `mcass_brsv` = EVA_Patients_5min_general_excluding_normal$`mcass_brsv`,
  `mcass_adrenergic_index` = EVA_Patients_5min_general_excluding_normal$`mcass_adrenergic_index`,
  `mcass_sudomotor_index` = EVA_Patients_5min_general_excluding_normal$`mcass_sudomotor_index`,
  `mcass_cardiovagal_index` = EVA_Patients_5min_general_excluding_normal$`mcass_cardiovagal_index`,
  `mcass_total` = EVA_Patients_5min_general_excluding_normal$`mcass_total`
  ),
  use = "pairwise.complete.obs", method = "spearman")

# Select variables for x and y axes
y_vars <- c("total_contractions_frequency", "total_contraction_rate_over10", "number_of_contractions_over26.7_per_hour", "number_of_total_epochs_frequency", "imputed_sum_of_amplitude", "motility_index")
x_vars <- c("cass_baseline.VM.SBP", "cass_baseline.VM.DBP", "cass_Valsalva.Ratio", "cass_HRDB.Ratio", "cass_adrenergic_index",  "cass_sudomotor_index", "cass_cardiovagal_index", "cass_total", "mcass_brsa", "mcass_brsv", "mcass_adrenergic_index", "mcass_sudomotor_index", "mcass_cardiovagal_index", "mcass_total")



# Subset the correlation matrix
sub_cor <- res[y_vars, x_vars]

# Convert to long format and plot with ggplot2
df_long <- melt(sub_cor)

# ggplot(df_long, aes(Var2, Var1, fill = value)) +
#   geom_tile(color = "white") +
#   geom_text(aes(label = round(value, 2)), size = 3) +
#   scale_fill_gradient2(
#     low = "red", mid = "white", high = "blue",
#     midpoint = 0, limit = c(-1, 1), name = "Correlation"
#   ) +
#   theme_minimal() +
#   theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust = 1)) +
#   coord_fixed()

# --- 1) helper: rectangular p-value matrix ---
get_pmat_rect <- function(df, y_vars, x_vars, method = "spearman", adjust = c("none","BH")) {
  adjust <- match.arg(adjust)
  p <- matrix(NA_real_, nrow = length(y_vars), ncol = length(x_vars),
              dimnames = list(y_vars, x_vars))
  for (i in seq_along(y_vars)) {
    for (j in seq_along(x_vars)) {
      x <- df[[x_vars[j]]]; y <- df[[y_vars[i]]]
      ok <- stats::complete.cases(x, y)
      if (sum(ok) >= 3) {
        p[i, j] <- suppressWarnings(cor.test(x[ok], y[ok], method = method)$p.value)
      }
    }
  }
  if (adjust != "none") {
    p[] <- p.adjust(as.vector(p), method = adjust)
  }
  p
}

# rectangular correlation submatrix
cor_mat <- cor(data.frame(
  total_contractions_frequency = EVA_Patients_5min_general_excluding_normal$total_contractions_frequency,
  total_contraction_rate_over10 = EVA_Patients_5min_general_excluding_normal$total_contraction_rate_over10,
  number_of_contractions_over26.7_per_hour = EVA_Patients_5min_general_excluding_normal$number_of_contractions_over26.7_per_hour,
  number_of_total_epochs_frequency = EVA_Patients_5min_general_excluding_normal$number_of_total_epochs_frequency,
  imputed_sum_of_amplitude = EVA_Patients_5min_general_excluding_normal$imputed_sum_of_amplitude,
  motility_index =EVA_Patients_5min_general_excluding_normal$motility_index,
  `cass_baseline VM SBP` = EVA_Patients_5min_general_excluding_normal$`cass_baseline VM SBP`, 
  `cass_baseline VM DBP` = EVA_Patients_5min_general_excluding_normal$`cass_baseline VM DBP`, 
  `cass_Valsalva Ratio` = EVA_Patients_5min_general_excluding_normal$`cass_Valsalva Ratio`, 
  `cass_HRDB Ratio` = EVA_Patients_5min_general_excluding_normal$`cass_HRDB Ratio`, 
  `cass_adrenergic_index` = EVA_Patients_5min_general_excluding_normal$`cass_adrenergic_index`, 
  `cass_sudomotor_index` = EVA_Patients_5min_general_excluding_normal$`cass_sudomotor_index`, 
  `cass_cardiovagal_index` = EVA_Patients_5min_general_excluding_normal$`cass_cardiovagal_index`, 
  `cass_total` = EVA_Patients_5min_general_excluding_normal$`cass_total`, 
  `mcass_brsa` = EVA_Patients_5min_general_excluding_normal$`mcass_brsa`, 
  `mcass_brsv` = EVA_Patients_5min_general_excluding_normal$`mcass_brsv`, 
  `mcass_adrenergic_index` = EVA_Patients_5min_general_excluding_normal$`mcass_adrenergic_index`, 
  `mcass_sudomotor_index` = EVA_Patients_5min_general_excluding_normal$`mcass_sudomotor_index`, 
  `mcass_cardiovagal_index` = EVA_Patients_5min_general_excluding_normal$`mcass_cardiovagal_index`, 
  `mcass_total` = EVA_Patients_5min_general_excluding_normal$`mcass_total`  
  ),
  use = "pairwise.complete.obs", method = "spearman")
sub_cor <- res[y_vars, x_vars]

# rectangular p-value matrix (choose method = "pearson"/"spearman"; adjust = "BH" if you want)
p_sub   <- get_pmat_rect(data.frame(
  total_contractions_frequency = EVA_Patients_5min_general_excluding_normal$total_contractions_frequency,
  total_contraction_rate_over10 = EVA_Patients_5min_general_excluding_normal$total_contraction_rate_over10,
  number_of_contractions_over26.7_per_hour = EVA_Patients_5min_general_excluding_normal$number_of_contractions_over26.7_per_hour,
  number_of_total_epochs_frequency = EVA_Patients_5min_general_excluding_normal$number_of_total_epochs_frequency,
  imputed_sum_of_amplitude = EVA_Patients_5min_general_excluding_normal$imputed_sum_of_amplitude,
  motility_index =EVA_Patients_5min_general_excluding_normal$motility_index,
  `cass_baseline VM SBP` = EVA_Patients_5min_general_excluding_normal$`cass_baseline VM SBP`, 
  `cass_baseline VM DBP` = EVA_Patients_5min_general_excluding_normal$`cass_baseline VM DBP`, 
  `cass_Valsalva Ratio` = EVA_Patients_5min_general_excluding_normal$`cass_Valsalva Ratio`, 
  `cass_HRDB Ratio` = EVA_Patients_5min_general_excluding_normal$`cass_HRDB Ratio`, 
  `cass_adrenergic_index` = EVA_Patients_5min_general_excluding_normal$`cass_adrenergic_index`, 
  `cass_sudomotor_index` = EVA_Patients_5min_general_excluding_normal$`cass_sudomotor_index`, 
  `cass_cardiovagal_index` = EVA_Patients_5min_general_excluding_normal$`cass_cardiovagal_index`, 
  `cass_total` = EVA_Patients_5min_general_excluding_normal$`cass_total`, 
  `mcass_brsa` = EVA_Patients_5min_general_excluding_normal$`mcass_brsa`, 
  `mcass_brsv` = EVA_Patients_5min_general_excluding_normal$`mcass_brsv`, 
  `mcass_adrenergic_index` = EVA_Patients_5min_general_excluding_normal$`mcass_adrenergic_index`, 
  `mcass_sudomotor_index` = EVA_Patients_5min_general_excluding_normal$`mcass_sudomotor_index`, 
  `mcass_cardiovagal_index` = EVA_Patients_5min_general_excluding_normal$`mcass_cardiovagal_index`, 
  `mcass_total` = EVA_Patients_5min_general_excluding_normal$`mcass_total`  
  ), y_vars, x_vars, method = "spearman", adjust = "none")

# long data for plotting
df_cor <- melt(sub_cor, varnames = c("Y","X"), value.name = "r")
df_p   <- melt(p_sub,   varnames = c("Y","X"), value.name = "p")

# flag significance (e.g., alpha = 0.05)
alpha <- 0.05
df_plot <- merge(df_cor, df_p, by = c("X","Y"))
df_plot$sig <- !is.na(df_plot$p) & df_plot$p < alpha

# --- 3) Plot A: correlation heatmap with numbers & significance marks ---
p_corr <- ggplot(df_plot, aes(X, Y, fill = r)) +
  geom_tile(color = "white") +
  geom_text(aes(label = sprintf("%.2f", r)), size = 3) +
  # add a star for significant cells
  # geom_text(data = subset(df_plot, sig), label = "*", vjust = -0.9, size = 4) +
  scale_fill_gradient2(low = "red", mid = "white", high = "blue",
                       midpoint = 0, limits = c(-1, 1), name = "Correlation") +
  coord_fixed() +
  theme_minimal(base_size = 12) +
  theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1),
        panel.grid = element_blank())

p_corr

# --- 4) Plot B: p-value heatmap with numbers (0–1 scale) ---
p_pval <- ggplot(df_plot, aes(X, Y, fill = p)) +
  geom_tile(color = "white") +
  geom_text(aes(label = ifelse(is.na(p), "", sprintf("%.3f", p))), size = 3) +
  scale_fill_gradient2(low = "red", high = "yellow",
                       limits = c(0, 1), name = "p-value") +
  coord_fixed() +
  theme_minimal(base_size = 12) +
  theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1),
        panel.grid = element_blank())

p_pval

EVA_Patients_5min_RMSSD_joined_excluding_normal <- EVA_Patients_5min_RMSSD_joined %>%
  filter(groups != "normal_SBTT_normal_CASS")

res <- cor(data.frame(
  `median_RMSSD_Time_fast epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`median_RMSSD_Time_fast epoch`,
  `median_RMSSD_Time_medium epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`median_RMSSD_Time_medium epoch`,
  `median_RMSSD_Time_slow epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`median_RMSSD_Time_slow epoch`,
  `median_RMSSD_Magnitude_fast epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`median_RMSSD_Magnitude_fast epoch`,
  `median_RMSSD_Magnitude_medium epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`median_RMSSD_Magnitude_medium epoch`,
  `median_RMSSD_Magnitude_slow epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`median_RMSSD_Magnitude_slow epoch`,
  `median_RMSSD_Time_total_epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$median_RMSSD_Time_total_epoch,
  `median_RMSSD_Magnitude_total_epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$median_RMSSD_Magnitude_total_epoch,
  `cass_baseline VM SBP` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_baseline VM SBP`,
  `cass_baseline VM DBP` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_baseline VM DBP`,
  `cass_Valsalva Ratio` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_Valsalva Ratio`,
  `cass_HRDB Ratio` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_HRDB Ratio`,
  `cass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_adrenergic_index`,
  `cass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_sudomotor_index`,
  `cass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_cardiovagal_index`,
  `cass_total` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_total`,
  `mcass_brsa` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`mcass_brsa`,
  `mcass_brsv` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`mcass_brsv`,
  `mcass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`mcass_adrenergic_index`,
  `mcass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`mcass_sudomotor_index`,
  `mcass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`mcass_cardiovagal_index`,
  `mcass_total` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`mcass_total`
  ),
  use = "pairwise.complete.obs", method = "spearman")

# Select variables for x and y axes
y_vars <- c("median_RMSSD_Magnitude_total_epoch", "median_RMSSD_Time_slow.epoch", "median_RMSSD_Time_medium.epoch", "median_RMSSD_Time_fast.epoch")
x_vars <- c("cass_baseline.VM.SBP", "cass_baseline.VM.DBP", "cass_Valsalva.Ratio", "cass_HRDB.Ratio", "cass_adrenergic_index",  "cass_sudomotor_index", "cass_cardiovagal_index", "cass_total", "mcass_brsa", "mcass_brsv", "mcass_adrenergic_index", "mcass_sudomotor_index", "mcass_cardiovagal_index", "mcass_total")

# Subset the correlation matrix
sub_cor <- res[y_vars, x_vars]

# Convert to long format and plot with ggplot2
df_long <- melt(sub_cor)
# 
# ggplot(df_long, aes(Var2, Var1, fill = value)) +
#   geom_tile(color = "white") +
#   geom_text(aes(label = round(value, 2)), size = 3) +
#   scale_fill_gradient2(
#     low = "red", mid = "white", high = "blue",
#     midpoint = 0, limit = c(-1, 1), name = "Correlation"
#   ) +
#   theme_minimal() +
#   theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust = 1)) +
#   coord_fixed()

# --- 1) helper: rectangular p-value matrix ---
get_pmat_rect <- function(df, y_vars, x_vars, method = "spearman", adjust = c("none","BH")) {
  adjust <- match.arg(adjust)
  p <- matrix(NA_real_, nrow = length(y_vars), ncol = length(x_vars),
              dimnames = list(y_vars, x_vars))
  for (i in seq_along(y_vars)) {
    for (j in seq_along(x_vars)) {
      x <- df[[x_vars[j]]]; y <- df[[y_vars[i]]]
      ok <- stats::complete.cases(x, y)
      if (sum(ok) >= 3) {
        p[i, j] <- suppressWarnings(cor.test(x[ok], y[ok], method = method)$p.value)
      }
    }
  }
  if (adjust != "none") {
    p[] <- p.adjust(as.vector(p), method = adjust)
  }
  p
}

# rectangular correlation submatrix
cor_mat <- cor(data.frame(
  `median_RMSSD_Time_fast epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`median_RMSSD_Time_fast epoch`, 
  `median_RMSSD_Time_medium epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`median_RMSSD_Time_medium epoch`, 
  `median_RMSSD_Time_slow epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`median_RMSSD_Time_slow epoch`, 
  `median_RMSSD_Magnitude_fast epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`median_RMSSD_Magnitude_fast epoch`, 
  `median_RMSSD_Magnitude_medium epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`median_RMSSD_Magnitude_medium epoch`, 
  `median_RMSSD_Magnitude_slow epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`median_RMSSD_Magnitude_slow epoch`, 
  `median_RMSSD_Time_total_epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$median_RMSSD_Time_total_epoch,
  `median_RMSSD_Magnitude_total_epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$median_RMSSD_Magnitude_total_epoch,
  `cass_baseline VM SBP` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_baseline VM SBP`, 
  `cass_baseline VM DBP` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_baseline VM DBP`, 
  `cass_Valsalva Ratio` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_Valsalva Ratio`, 
  `cass_HRDB Ratio` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_HRDB Ratio`, 
  `cass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_adrenergic_index`, 
  `cass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_sudomotor_index`, 
  `cass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_cardiovagal_index`, 
  `cass_total` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_total`, 
  `mcass_brsa` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`mcass_brsa`, 
  `mcass_brsv` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`mcass_brsv`, 
  `mcass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`mcass_adrenergic_index`, 
  `mcass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`mcass_sudomotor_index`, 
  `mcass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`mcass_cardiovagal_index`, 
  `mcass_total` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`mcass_total`
  ),
  use = "pairwise.complete.obs", method = "spearman")
sub_cor <- res[y_vars, x_vars]

# rectangular p-value matrix
p_sub   <- get_pmat_rect(data.frame(
  `median_RMSSD_Time_fast epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`median_RMSSD_Time_fast epoch`, 
  `median_RMSSD_Time_medium epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`median_RMSSD_Time_medium epoch`, 
  `median_RMSSD_Time_slow epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`median_RMSSD_Time_slow epoch`, 
  `median_RMSSD_Magnitude_fast epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`median_RMSSD_Magnitude_fast epoch`, 
  `median_RMSSD_Magnitude_medium epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`median_RMSSD_Magnitude_medium epoch`, 
  `median_RMSSD_Magnitude_slow epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`median_RMSSD_Magnitude_slow epoch`, 
  `median_RMSSD_Time_total_epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$median_RMSSD_Time_total_epoch,
  `median_RMSSD_Magnitude_total_epoch` = EVA_Patients_5min_RMSSD_joined_excluding_normal$median_RMSSD_Magnitude_total_epoch,
  `cass_baseline VM SBP` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_baseline VM SBP`, 
  `cass_baseline VM DBP` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_baseline VM DBP`, 
  `cass_Valsalva Ratio` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_Valsalva Ratio`, 
  `cass_HRDB Ratio` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_HRDB Ratio`, 
  `cass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_adrenergic_index`, 
  `cass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_sudomotor_index`, 
  `cass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_cardiovagal_index`, 
  `cass_total` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`cass_total`, 
  `mcass_brsa` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`mcass_brsa`, 
  `mcass_brsv` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`mcass_brsv`, 
  `mcass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`mcass_adrenergic_index`, 
  `mcass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`mcass_sudomotor_index`, 
  `mcass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`mcass_cardiovagal_index`, 
  `mcass_total` = EVA_Patients_5min_RMSSD_joined_excluding_normal$`mcass_total`
  ), y_vars, x_vars, method = "spearman", adjust = "none")

# long data for plotting
df_cor <- melt(sub_cor, varnames = c("Y","X"), value.name = "r")
df_p   <- melt(p_sub,   varnames = c("Y","X"), value.name = "p")

# flag significance (e.g., alpha = 0.05)
alpha <- 0.05
df_plot <- merge(df_cor, df_p, by = c("X","Y"))
df_plot$sig <- !is.na(df_plot$p) & df_plot$p < alpha

# --- 3) Plot A: correlation heatmap with numbers & significance marks ---
p_corr <- ggplot(df_plot, aes(X, Y, fill = r)) +
  geom_tile(color = "white") +
  geom_text(aes(label = sprintf("%.2f", r)), size = 3) +
  # add a star for significant cells
  # geom_text(data = subset(df_plot, sig), label = "*", vjust = -0.9, size = 4) +
  scale_fill_gradient2(low = "red", mid = "white", high = "blue",
                       midpoint = 0, limits = c(-1, 1), name = "Correlation") +
  coord_fixed() +
  theme_minimal(base_size = 12) +
  theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1),
        panel.grid = element_blank())

p_corr

# --- 4) Plot B: p-value heatmap with numbers (0–1 scale) ---
p_pval <- ggplot(df_plot, aes(X, Y, fill = p)) +
  geom_tile(color = "white") +
  geom_text(aes(label = ifelse(is.na(p), "", sprintf("%.3f", p))), size = 3) +
  scale_fill_gradient2(low = "red", high = "yellow",
                       limits = c(0, 1), name = "p-value") +
  coord_fixed() +
  theme_minimal(base_size = 12) +
  theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1),
        panel.grid = element_blank())

p_pval

5.8 New Correlation Matrix 3

res <- cor(data.frame(
  total_contractions_frequency = EVA_Patients_5min_general$total_contractions_frequency,
  total_contraction_rate_over10 = EVA_Patients_5min_general$total_contraction_rate_over10,
  number_of_contractions_over26.7_per_hour = EVA_Patients_5min_general$number_of_contractions_over26.7_per_hour,
  number_of_total_epochs_frequency = EVA_Patients_5min_general$number_of_total_epochs_frequency,
  imputed_sum_of_amplitude = EVA_Patients_5min_general$imputed_sum_of_amplitude,
  motility_index =EVA_Patients_5min_general$motility_index,
  `cass_baseline VM SBP` = EVA_Patients_5min_general$`cass_baseline VM SBP`,
  `cass_baseline VM DBP` = EVA_Patients_5min_general$`cass_baseline VM DBP`,
  `cass_Valsalva Ratio` = EVA_Patients_5min_general$`cass_Valsalva Ratio`,
  `cass_HRDB Ratio` = EVA_Patients_5min_general$`cass_HRDB Ratio`,
  `cass_adrenergic_index` = EVA_Patients_5min_general$`cass_adrenergic_index`,
  `cass_sudomotor_index` = EVA_Patients_5min_general$`cass_sudomotor_index`,
  `cass_cardiovagal_index` = EVA_Patients_5min_general$`cass_cardiovagal_index`,
  `cass_total` = EVA_Patients_5min_general$`cass_total`,
  `mcass_brsa` = EVA_Patients_5min_general$`mcass_brsa`,
  `mcass_brsv` = EVA_Patients_5min_general$`mcass_brsv`,
  `mcass_adrenergic_index` = EVA_Patients_5min_general$`mcass_adrenergic_index`,
  `mcass_sudomotor_index` = EVA_Patients_5min_general$`mcass_sudomotor_index`,
  `mcass_cardiovagal_index` = EVA_Patients_5min_general$`mcass_cardiovagal_index`,
  `mcass_total` = EVA_Patients_5min_general$`mcass_total`
  ),
  use = "pairwise.complete.obs", method = "spearman")

# Select variables for x and y axes
y_vars <- c("total_contractions_frequency", "total_contraction_rate_over10", "number_of_contractions_over26.7_per_hour", "number_of_total_epochs_frequency", "imputed_sum_of_amplitude", "motility_index")
x_vars <- c("cass_baseline.VM.SBP", "cass_baseline.VM.DBP", "cass_Valsalva.Ratio", "cass_HRDB.Ratio", "cass_adrenergic_index",  "cass_sudomotor_index", "cass_cardiovagal_index", "cass_total", "mcass_brsa", "mcass_brsv", "mcass_adrenergic_index", "mcass_sudomotor_index", "mcass_cardiovagal_index", "mcass_total")



# Subset the correlation matrix
sub_cor <- res[y_vars, x_vars]

# Convert to long format and plot with ggplot2
df_long <- melt(sub_cor)
# 
# ggplot(df_long, aes(Var2, Var1, fill = value)) +
#   geom_tile(color = "white") +
#   geom_text(aes(label = round(value, 2)), size = 3) +
#   scale_fill_gradient2(
#     low = "red", mid = "white", high = "blue",
#     midpoint = 0, limit = c(-1, 1), name = "Correlation"
#   ) +
#   theme_minimal() +
#   theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust = 1)) +
#   coord_fixed()

# --- 1) helper: rectangular p-value matrix ---
get_pmat_rect <- function(df, y_vars, x_vars, method = "spearman", adjust = c("none","BH")) {
  adjust <- match.arg(adjust)
  p <- matrix(NA_real_, nrow = length(y_vars), ncol = length(x_vars),
              dimnames = list(y_vars, x_vars))
  for (i in seq_along(y_vars)) {
    for (j in seq_along(x_vars)) {
      x <- df[[x_vars[j]]]; y <- df[[y_vars[i]]]
      ok <- stats::complete.cases(x, y)
      if (sum(ok) >= 3) {
        p[i, j] <- suppressWarnings(cor.test(x[ok], y[ok], method = method)$p.value)
      }
    }
  }
  if (adjust != "none") {
    p[] <- p.adjust(as.vector(p), method = adjust)
  }
  p
}

# rectangular correlation submatrix
cor_mat <- cor(data.frame(
  total_contractions_frequency = EVA_Patients_5min_general$total_contractions_frequency,
  total_contraction_rate_over10 = EVA_Patients_5min_general$total_contraction_rate_over10,
  number_of_contractions_over26.7_per_hour = EVA_Patients_5min_general$number_of_contractions_over26.7_per_hour,
  number_of_total_epochs_frequency = EVA_Patients_5min_general$number_of_total_epochs_frequency,
  imputed_sum_of_amplitude = EVA_Patients_5min_general$imputed_sum_of_amplitude,
  motility_index =EVA_Patients_5min_general$motility_index,
  `cass_baseline VM SBP` = EVA_Patients_5min_general$`cass_baseline VM SBP`, 
  `cass_baseline VM DBP` = EVA_Patients_5min_general$`cass_baseline VM DBP`, 
  `cass_Valsalva Ratio` = EVA_Patients_5min_general$`cass_Valsalva Ratio`, 
  `cass_HRDB Ratio` = EVA_Patients_5min_general$`cass_HRDB Ratio`, 
  `cass_adrenergic_index` = EVA_Patients_5min_general$`cass_adrenergic_index`, 
  `cass_sudomotor_index` = EVA_Patients_5min_general$`cass_sudomotor_index`, 
  `cass_cardiovagal_index` = EVA_Patients_5min_general$`cass_cardiovagal_index`, 
  `cass_total` = EVA_Patients_5min_general$`cass_total`, 
  `mcass_brsa` = EVA_Patients_5min_general$`mcass_brsa`, 
  `mcass_brsv` = EVA_Patients_5min_general$`mcass_brsv`, 
  `mcass_adrenergic_index` = EVA_Patients_5min_general$`mcass_adrenergic_index`, 
  `mcass_sudomotor_index` = EVA_Patients_5min_general$`mcass_sudomotor_index`, 
  `mcass_cardiovagal_index` = EVA_Patients_5min_general$`mcass_cardiovagal_index`, 
  `mcass_total` = EVA_Patients_5min_general$`mcass_total`  
  ),
  use = "pairwise.complete.obs", method = "spearman")
sub_cor <- res[y_vars, x_vars]

# rectangular p-value matrix (choose method = "pearson"/"spearman"; adjust = "BH" if you want)
p_sub   <- get_pmat_rect(data.frame(
  total_contractions_frequency = EVA_Patients_5min_general$total_contractions_frequency,
  total_contraction_rate_over10 = EVA_Patients_5min_general$total_contraction_rate_over10,
  number_of_contractions_over26.7_per_hour = EVA_Patients_5min_general$number_of_contractions_over26.7_per_hour,
  number_of_total_epochs_frequency = EVA_Patients_5min_general$number_of_total_epochs_frequency,
  imputed_sum_of_amplitude = EVA_Patients_5min_general$imputed_sum_of_amplitude,
  motility_index =EVA_Patients_5min_general$motility_index,
  `cass_baseline VM SBP` = EVA_Patients_5min_general$`cass_baseline VM SBP`, 
  `cass_baseline VM DBP` = EVA_Patients_5min_general$`cass_baseline VM DBP`, 
  `cass_Valsalva Ratio` = EVA_Patients_5min_general$`cass_Valsalva Ratio`, 
  `cass_HRDB Ratio` = EVA_Patients_5min_general$`cass_HRDB Ratio`, 
  `cass_adrenergic_index` = EVA_Patients_5min_general$`cass_adrenergic_index`, 
  `cass_sudomotor_index` = EVA_Patients_5min_general$`cass_sudomotor_index`, 
  `cass_cardiovagal_index` = EVA_Patients_5min_general$`cass_cardiovagal_index`, 
  `cass_total` = EVA_Patients_5min_general$`cass_total`, 
  `mcass_brsa` = EVA_Patients_5min_general$`mcass_brsa`, 
  `mcass_brsv` = EVA_Patients_5min_general$`mcass_brsv`, 
  `mcass_adrenergic_index` = EVA_Patients_5min_general$`mcass_adrenergic_index`, 
  `mcass_sudomotor_index` = EVA_Patients_5min_general$`mcass_sudomotor_index`, 
  `mcass_cardiovagal_index` = EVA_Patients_5min_general$`mcass_cardiovagal_index`, 
  `mcass_total` = EVA_Patients_5min_general$`mcass_total`  
  ), y_vars, x_vars, method = "spearman", adjust = "none")

# long data for plotting
df_cor <- melt(sub_cor, varnames = c("Y","X"), value.name = "r")
df_p   <- melt(p_sub,   varnames = c("Y","X"), value.name = "p")

# flag significance (e.g., alpha = 0.05)
alpha <- 0.05
df_plot <- merge(df_cor, df_p, by = c("X","Y"))
df_plot$sig <- !is.na(df_plot$p) & df_plot$p < alpha

# --- 3) Plot A: correlation heatmap with numbers & significance marks ---
p_corr <- ggplot(df_plot, aes(X, Y, fill = r)) +
  geom_tile(color = "white") +
  geom_text(aes(label = sprintf("%.2f", r)), size = 3) +
  # add a star for significant cells
  # geom_text(data = subset(df_plot, sig), label = "*", vjust = -0.9, size = 4) +
  scale_fill_gradient2(low = "red", mid = "white", high = "blue",
                       midpoint = 0, limits = c(-1, 1), name = "Correlation") +
  coord_fixed() +
  theme_minimal(base_size = 12) +
  theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1),
        panel.grid = element_blank())

p_corr

# --- 4) Plot B: p-value heatmap with numbers (0–1 scale) ---
p_pval <- ggplot(df_plot, aes(X, Y, fill = p)) +
  geom_tile(color = "white") +
  geom_text(aes(label = ifelse(is.na(p), "", sprintf("%.3f", p))), size = 3) +
  scale_fill_gradient2(low = "red", high = "yellow",
                       limits = c(0, 1), name = "p-value") +
  coord_fixed() +
  theme_minimal(base_size = 12) +
  theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1),
        panel.grid = element_blank())

p_pval

res <- cor(data.frame(
  `median_RMSSD_Time_fast epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_fast epoch`,
  `median_RMSSD_Time_medium epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_medium epoch`,
  `median_RMSSD_Time_slow epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_slow epoch`,
  `median_RMSSD_Magnitude_fast epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_fast epoch`,
  `median_RMSSD_Magnitude_medium epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_medium epoch`,
  `median_RMSSD_Magnitude_slow epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_slow epoch`,
  `median_RMSSD_Time_total_epoch` = EVA_Patients_5min_RMSSD_joined$median_RMSSD_Time_total_epoch,
  `median_RMSSD_Magnitude_total_epoch` = EVA_Patients_5min_RMSSD_joined$median_RMSSD_Magnitude_total_epoch,
  `cass_baseline VM SBP` = EVA_Patients_5min_RMSSD_joined$`cass_baseline VM SBP`,
  `cass_baseline VM DBP` = EVA_Patients_5min_RMSSD_joined$`cass_baseline VM DBP`,
  `cass_Valsalva Ratio` = EVA_Patients_5min_RMSSD_joined$`cass_Valsalva Ratio`,
  `cass_HRDB Ratio` = EVA_Patients_5min_RMSSD_joined$`cass_HRDB Ratio`,
  `cass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined$`cass_adrenergic_index`,
  `cass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined$`cass_sudomotor_index`,
  `cass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined$`cass_cardiovagal_index`,
  `cass_total` = EVA_Patients_5min_RMSSD_joined$`cass_total`,
  `mcass_brsa` = EVA_Patients_5min_RMSSD_joined$`mcass_brsa`,
  `mcass_brsv` = EVA_Patients_5min_RMSSD_joined$`mcass_brsv`,
  `mcass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined$`mcass_adrenergic_index`,
  `mcass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined$`mcass_sudomotor_index`,
  `mcass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined$`mcass_cardiovagal_index`,
  `mcass_total` = EVA_Patients_5min_RMSSD_joined$`mcass_total`
  ),
  use = "pairwise.complete.obs", method = "spearman")

# Select variables for x and y axes
y_vars <- c("median_RMSSD_Magnitude_total_epoch", "median_RMSSD_Time_slow.epoch", "median_RMSSD_Time_medium.epoch", "median_RMSSD_Time_fast.epoch")
x_vars <- c("cass_baseline.VM.SBP", "cass_baseline.VM.DBP", "cass_Valsalva.Ratio", "cass_HRDB.Ratio", "cass_adrenergic_index",  "cass_sudomotor_index", "cass_cardiovagal_index", "cass_total", "mcass_brsa", "mcass_brsv", "mcass_adrenergic_index", "mcass_sudomotor_index", "mcass_cardiovagal_index", "mcass_total")

# Subset the correlation matrix
sub_cor <- res[y_vars, x_vars]
# 
# # Convert to long format and plot with ggplot2
# df_long <- melt(sub_cor)
# 
# ggplot(df_long, aes(Var2, Var1, fill = value)) +
#   geom_tile(color = "white") +
#   geom_text(aes(label = round(value, 2)), size = 3) +
#   scale_fill_gradient2(
#     low = "red", mid = "white", high = "blue",
#     midpoint = 0, limit = c(-1, 1), name = "Correlation"
#   ) +
#   theme_minimal() +
#   theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust = 1)) +
#   coord_fixed()

# --- 1) helper: rectangular p-value matrix ---
get_pmat_rect <- function(df, y_vars, x_vars, method = "spearman", adjust = c("none","BH")) {
  adjust <- match.arg(adjust)
  p <- matrix(NA_real_, nrow = length(y_vars), ncol = length(x_vars),
              dimnames = list(y_vars, x_vars))
  for (i in seq_along(y_vars)) {
    for (j in seq_along(x_vars)) {
      x <- df[[x_vars[j]]]; y <- df[[y_vars[i]]]
      ok <- stats::complete.cases(x, y)
      if (sum(ok) >= 3) {
        p[i, j] <- suppressWarnings(cor.test(x[ok], y[ok], method = method)$p.value)
      }
    }
  }
  if (adjust != "none") {
    p[] <- p.adjust(as.vector(p), method = adjust)
  }
  p
}

# rectangular correlation submatrix
cor_mat <- cor(data.frame(
  `median_RMSSD_Time_fast epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_fast epoch`, 
  `median_RMSSD_Time_medium epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_medium epoch`, 
  `median_RMSSD_Time_slow epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_slow epoch`, 
  `median_RMSSD_Magnitude_fast epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_fast epoch`, 
  `median_RMSSD_Magnitude_medium epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_medium epoch`, 
  `median_RMSSD_Magnitude_slow epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_slow epoch`, 
  `median_RMSSD_Time_total_epoch` = EVA_Patients_5min_RMSSD_joined$median_RMSSD_Time_total_epoch,
  `median_RMSSD_Magnitude_total_epoch` = EVA_Patients_5min_RMSSD_joined$median_RMSSD_Magnitude_total_epoch,
  `cass_baseline VM SBP` = EVA_Patients_5min_RMSSD_joined$`cass_baseline VM SBP`, 
  `cass_baseline VM DBP` = EVA_Patients_5min_RMSSD_joined$`cass_baseline VM DBP`, 
  `cass_Valsalva Ratio` = EVA_Patients_5min_RMSSD_joined$`cass_Valsalva Ratio`, 
  `cass_HRDB Ratio` = EVA_Patients_5min_RMSSD_joined$`cass_HRDB Ratio`, 
  `cass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined$`cass_adrenergic_index`, 
  `cass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined$`cass_sudomotor_index`, 
  `cass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined$`cass_cardiovagal_index`, 
  `cass_total` = EVA_Patients_5min_RMSSD_joined$`cass_total`, 
  `mcass_brsa` = EVA_Patients_5min_RMSSD_joined$`mcass_brsa`, 
  `mcass_brsv` = EVA_Patients_5min_RMSSD_joined$`mcass_brsv`, 
  `mcass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined$`mcass_adrenergic_index`, 
  `mcass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined$`mcass_sudomotor_index`, 
  `mcass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined$`mcass_cardiovagal_index`, 
  `mcass_total` = EVA_Patients_5min_RMSSD_joined$`mcass_total`
  ),
  use = "pairwise.complete.obs", method = "spearman")
sub_cor <- res[y_vars, x_vars]

# rectangular p-value matrix
p_sub   <- get_pmat_rect(data.frame(
  `median_RMSSD_Time_fast epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_fast epoch`, 
  `median_RMSSD_Time_medium epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_medium epoch`, 
  `median_RMSSD_Time_slow epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_slow epoch`, 
  `median_RMSSD_Magnitude_fast epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_fast epoch`, 
  `median_RMSSD_Magnitude_medium epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_medium epoch`, 
  `median_RMSSD_Magnitude_slow epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_slow epoch`, 
  `median_RMSSD_Time_total_epoch` = EVA_Patients_5min_RMSSD_joined$median_RMSSD_Time_total_epoch,
  `median_RMSSD_Magnitude_total_epoch` = EVA_Patients_5min_RMSSD_joined$median_RMSSD_Magnitude_total_epoch,
  `cass_baseline VM SBP` = EVA_Patients_5min_RMSSD_joined$`cass_baseline VM SBP`, 
  `cass_baseline VM DBP` = EVA_Patients_5min_RMSSD_joined$`cass_baseline VM DBP`, 
  `cass_Valsalva Ratio` = EVA_Patients_5min_RMSSD_joined$`cass_Valsalva Ratio`, 
  `cass_HRDB Ratio` = EVA_Patients_5min_RMSSD_joined$`cass_HRDB Ratio`, 
  `cass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined$`cass_adrenergic_index`, 
  `cass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined$`cass_sudomotor_index`, 
  `cass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined$`cass_cardiovagal_index`, 
  `cass_total` = EVA_Patients_5min_RMSSD_joined$`cass_total`, 
  `mcass_brsa` = EVA_Patients_5min_RMSSD_joined$`mcass_brsa`, 
  `mcass_brsv` = EVA_Patients_5min_RMSSD_joined$`mcass_brsv`, 
  `mcass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined$`mcass_adrenergic_index`, 
  `mcass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined$`mcass_sudomotor_index`, 
  `mcass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined$`mcass_cardiovagal_index`, 
  `mcass_total` = EVA_Patients_5min_RMSSD_joined$`mcass_total`
  ), y_vars, x_vars, method = "spearman", adjust = "none")

# long data for plotting
df_cor <- melt(sub_cor, varnames = c("Y","X"), value.name = "r")
df_p   <- melt(p_sub,   varnames = c("Y","X"), value.name = "p")

# flag significance (e.g., alpha = 0.05)
alpha <- 0.05
df_plot <- merge(df_cor, df_p, by = c("X","Y"))
df_plot$sig <- !is.na(df_plot$p) & df_plot$p < alpha

# --- 3) Plot A: correlation heatmap with numbers & significance marks ---
p_corr <- ggplot(df_plot, aes(X, Y, fill = r)) +
  geom_tile(color = "white") +
  geom_text(aes(label = sprintf("%.2f", r)), size = 3) +
  # add a star for significant cells
  # geom_text(data = subset(df_plot, sig), label = "*", vjust = -0.9, size = 4) +
  scale_fill_gradient2(low = "red", mid = "white", high = "blue",
                       midpoint = 0, limits = c(-1, 1), name = "Correlation") +
  coord_fixed() +
  theme_minimal(base_size = 12) +
  theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1),
        panel.grid = element_blank())

p_corr

# --- 4) Plot B: p-value heatmap with numbers (0–1 scale) ---
p_pval <- ggplot(df_plot, aes(X, Y, fill = p)) +
  geom_tile(color = "white") +
  geom_text(aes(label = ifelse(is.na(p), "", sprintf("%.3f", p))), size = 3) +
  scale_fill_gradient2(low = "red", high = "yellow",
                       limits = c(0, 1), name = "p-value") +
  coord_fixed() +
  theme_minimal(base_size = 12) +
  theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1),
        panel.grid = element_blank())

p_pval

res_new <- cor(data.frame(
  total_contractions_frequency = EVA_Patients_5min_general$total_contractions_frequency,
  total_contraction_rate_over10 = EVA_Patients_5min_general$total_contraction_rate_over10,
  number_of_contractions_over26.7_per_hour = EVA_Patients_5min_general$number_of_contractions_over26.7_per_hour,
  number_of_total_epochs_frequency = EVA_Patients_5min_general$number_of_total_epochs_frequency,
  imputed_sum_of_amplitude = EVA_Patients_5min_general$imputed_sum_of_amplitude,
  motility_index =EVA_Patients_5min_general$motility_index,
  `mcass_brsa` = EVA_Patients_5min_general$`mcass_brsa`,
  `mcass_brsv` = EVA_Patients_5min_general$`mcass_brsv`,
  `mcass_adrenergic_index` = EVA_Patients_5min_general$`mcass_adrenergic_index`,
  `mcass_sudomotor_index` = EVA_Patients_5min_general$`mcass_sudomotor_index`,
  `mcass_cardiovagal_index` = EVA_Patients_5min_general$`mcass_cardiovagal_index`,
  `mcass_total` = EVA_Patients_5min_general$`mcass_total`
  ),
  use = "pairwise.complete.obs", method = "spearman")

# Select variables for x and y axes
y_vars <- c("total_contractions_frequency", "total_contraction_rate_over10", "number_of_contractions_over26.7_per_hour", "number_of_total_epochs_frequency", "imputed_sum_of_amplitude", "motility_index")
x_vars_new <- c("mcass_brsa", "mcass_brsv", "mcass_adrenergic_index", "mcass_sudomotor_index", "mcass_cardiovagal_index", "mcass_total")



# Subset the correlation matrix
sub_cor <- res_new[y_vars, x_vars_new]

# Convert to long format and plot with ggplot2
df_long <- melt(sub_cor)
# 
# ggplot(df_long, aes(Var2, Var1, fill = value)) +
#   geom_tile(color = "white") +
#   geom_text(aes(label = round(value, 2)), size = 3) +
#   scale_fill_gradient2(
#     low = "red", mid = "white", high = "blue",
#     midpoint = 0, limit = c(-1, 1), name = "Correlation"
#   ) +
#   theme_minimal() +
#   theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust = 1)) +
#   coord_fixed()

# --- 1) helper: rectangular p-value matrix ---
get_pmat_rect <- function(df, y_vars, x_vars, method = "spearman", adjust = c("none","BH")) {
  adjust <- match.arg(adjust)
  p <- matrix(NA_real_, nrow = length(y_vars), ncol = length(x_vars),
              dimnames = list(y_vars, x_vars))
  for (i in seq_along(y_vars)) {
    for (j in seq_along(x_vars)) {
      x <- df[[x_vars[j]]]; y <- df[[y_vars[i]]]
      ok <- stats::complete.cases(x, y)
      if (sum(ok) >= 3) {
        p[i, j] <- suppressWarnings(cor.test(x[ok], y[ok], method = method)$p.value)
      }
    }
  }
  if (adjust != "none") {
    p[] <- p.adjust(as.vector(p), method = adjust)
  }
  p
}

# rectangular correlation submatrix
cor_mat <- cor(data.frame(
  total_contractions_frequency = EVA_Patients_5min_general$total_contractions_frequency,
  total_contraction_rate_over10 = EVA_Patients_5min_general$total_contraction_rate_over10,
  number_of_contractions_over26.7_per_hour = EVA_Patients_5min_general$number_of_contractions_over26.7_per_hour,
  number_of_total_epochs_frequency = EVA_Patients_5min_general$number_of_total_epochs_frequency,
  imputed_sum_of_amplitude = EVA_Patients_5min_general$imputed_sum_of_amplitude,
  motility_index =EVA_Patients_5min_general$motility_index,
  `mcass_brsa` = EVA_Patients_5min_general$`mcass_brsa`, 
  `mcass_brsv` = EVA_Patients_5min_general$`mcass_brsv`, 
  `mcass_adrenergic_index` = EVA_Patients_5min_general$`mcass_adrenergic_index`, 
  `mcass_sudomotor_index` = EVA_Patients_5min_general$`mcass_sudomotor_index`, 
  `mcass_cardiovagal_index` = EVA_Patients_5min_general$`mcass_cardiovagal_index`, 
  `mcass_total` = EVA_Patients_5min_general$`mcass_total`  
  ),
  use = "pairwise.complete.obs", method = "spearman")
sub_cor <- res_new[y_vars, x_vars_new]

# rectangular p-value matrix (choose method = "pearson"/"spearman"; adjust = "BH" if you want)
p_sub   <- get_pmat_rect(data.frame(
  total_contractions_frequency = EVA_Patients_5min_general$total_contractions_frequency,
  total_contraction_rate_over10 = EVA_Patients_5min_general$total_contraction_rate_over10,
  number_of_contractions_over26.7_per_hour = EVA_Patients_5min_general$number_of_contractions_over26.7_per_hour,
  number_of_total_epochs_frequency = EVA_Patients_5min_general$number_of_total_epochs_frequency,
  imputed_sum_of_amplitude = EVA_Patients_5min_general$imputed_sum_of_amplitude,
  motility_index =EVA_Patients_5min_general$motility_index,
  `mcass_brsa` = EVA_Patients_5min_general$`mcass_brsa`, 
  `mcass_brsv` = EVA_Patients_5min_general$`mcass_brsv`, 
  `mcass_adrenergic_index` = EVA_Patients_5min_general$`mcass_adrenergic_index`, 
  `mcass_sudomotor_index` = EVA_Patients_5min_general$`mcass_sudomotor_index`, 
  `mcass_cardiovagal_index` = EVA_Patients_5min_general$`mcass_cardiovagal_index`, 
  `mcass_total` = EVA_Patients_5min_general$`mcass_total`  
  ), y_vars, x_vars_new, method = "spearman", adjust = "none")

# long data for plotting
df_cor <- melt(sub_cor, varnames = c("Y","X"), value.name = "r")
df_p   <- melt(p_sub,   varnames = c("Y","X"), value.name = "p")

# flag significance (e.g., alpha = 0.05)
alpha <- 0.05
df_plot <- merge(df_cor, df_p, by = c("X","Y"))
df_plot$sig <- !is.na(df_plot$p) & df_plot$p < alpha

# --- 3) Plot A: correlation heatmap with numbers & significance marks ---
p_corr <- ggplot(df_plot, aes(X, Y, fill = r)) +
  geom_tile(color = "white") +
  geom_text(aes(label = sprintf("%.2f", r)), size = 3) +
  # add a star for significant cells
  # geom_text(data = subset(df_plot, sig), label = "*", vjust = -0.9, size = 4) +
  scale_fill_gradient2(low = "red", mid = "white", high = "blue",
                       midpoint = 0, limits = c(-1, 1), name = "Correlation") +
  coord_fixed() +
  theme_minimal(base_size = 12) +
  theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1),
        panel.grid = element_blank())

p_corr

# --- 4) Plot B: p-value heatmap with numbers (0–1 scale) ---
p_pval <- ggplot(df_plot, aes(X, Y, fill = p)) +
  geom_tile(color = "white") +
  geom_text(aes(label = ifelse(is.na(p), "", sprintf("%.3f", p))), size = 3) +
  scale_fill_gradient2(low = "red", high = "yellow",
                       limits = c(0, 1), name = "p-value") +
  coord_fixed() +
  theme_minimal(base_size = 12) +
  theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1),
        panel.grid = element_blank())

p_pval

res_new <- cor(data.frame(
  `median_RMSSD_Time_fast epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_fast epoch`,
  `median_RMSSD_Time_medium epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_medium epoch`,
  `median_RMSSD_Time_slow epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_slow epoch`,
  `median_RMSSD_Magnitude_fast epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_fast epoch`,
  `median_RMSSD_Magnitude_medium epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_medium epoch`,
  `median_RMSSD_Magnitude_slow epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_slow epoch`,
  `median_RMSSD_Time_total_epoch` = EVA_Patients_5min_RMSSD_joined$median_RMSSD_Time_total_epoch,
  `median_RMSSD_Magnitude_total_epoch` = EVA_Patients_5min_RMSSD_joined$median_RMSSD_Magnitude_total_epoch,
  `mcass_brsa` = EVA_Patients_5min_RMSSD_joined$`mcass_brsa`,
  `mcass_brsv` = EVA_Patients_5min_RMSSD_joined$`mcass_brsv`,
  `mcass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined$`mcass_adrenergic_index`,
  `mcass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined$`mcass_sudomotor_index`,
  `mcass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined$`mcass_cardiovagal_index`,
  `mcass_total` = EVA_Patients_5min_RMSSD_joined$`mcass_total`
  ),
  use = "pairwise.complete.obs", method = "spearman")

# Select variables for x and y axes
y_vars <- c("median_RMSSD_Magnitude_total_epoch", "median_RMSSD_Time_slow.epoch", "median_RMSSD_Time_medium.epoch", "median_RMSSD_Time_fast.epoch")
x_vars_new <- c("mcass_brsa", "mcass_brsv", "mcass_adrenergic_index", "mcass_sudomotor_index", "mcass_cardiovagal_index", "mcass_total")

# Subset the correlation matrix
sub_cor <- res_new[y_vars, x_vars_new]
# 
# # Convert to long format and plot with ggplot2
# df_long <- melt(sub_cor)
# 
# ggplot(df_long, aes(Var2, Var1, fill = value)) +
#   geom_tile(color = "white") +
#   geom_text(aes(label = round(value, 2)), size = 3) +
#   scale_fill_gradient2(
#     low = "red", mid = "white", high = "blue",
#     midpoint = 0, limit = c(-1, 1), name = "Correlation"
#   ) +
#   theme_minimal() +
#   theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust = 1)) +
#   coord_fixed()

# --- 1) helper: rectangular p-value matrix ---
get_pmat_rect <- function(df, y_vars, x_vars, method = "spearman", adjust = c("none","BH")) {
  adjust <- match.arg(adjust)
  p <- matrix(NA_real_, nrow = length(y_vars), ncol = length(x_vars),
              dimnames = list(y_vars, x_vars))
  for (i in seq_along(y_vars)) {
    for (j in seq_along(x_vars)) {
      x <- df[[x_vars[j]]]; y <- df[[y_vars[i]]]
      ok <- stats::complete.cases(x, y)
      if (sum(ok) >= 3) {
        p[i, j] <- suppressWarnings(cor.test(x[ok], y[ok], method = method)$p.value)
      }
    }
  }
  if (adjust != "none") {
    p[] <- p.adjust(as.vector(p), method = adjust)
  }
  p
}

# rectangular correlation submatrix
cor_mat <- cor(data.frame(
  `median_RMSSD_Time_fast epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_fast epoch`, 
  `median_RMSSD_Time_medium epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_medium epoch`, 
  `median_RMSSD_Time_slow epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_slow epoch`, 
  `median_RMSSD_Magnitude_fast epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_fast epoch`, 
  `median_RMSSD_Magnitude_medium epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_medium epoch`, 
  `median_RMSSD_Magnitude_slow epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_slow epoch`, 
  `median_RMSSD_Time_total_epoch` = EVA_Patients_5min_RMSSD_joined$median_RMSSD_Time_total_epoch,
  `median_RMSSD_Magnitude_total_epoch` = EVA_Patients_5min_RMSSD_joined$median_RMSSD_Magnitude_total_epoch,
  `mcass_brsa` = EVA_Patients_5min_RMSSD_joined$`mcass_brsa`, 
  `mcass_brsv` = EVA_Patients_5min_RMSSD_joined$`mcass_brsv`, 
  `mcass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined$`mcass_adrenergic_index`, 
  `mcass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined$`mcass_sudomotor_index`, 
  `mcass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined$`mcass_cardiovagal_index`, 
  `mcass_total` = EVA_Patients_5min_RMSSD_joined$`mcass_total`
  ),
  use = "pairwise.complete.obs", method = "spearman")
sub_cor_new <- res_new[y_vars, x_vars_new]

# rectangular p-value matrix
p_sub   <- get_pmat_rect(data.frame(
  `median_RMSSD_Time_fast epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_fast epoch`, 
  `median_RMSSD_Time_medium epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_medium epoch`, 
  `median_RMSSD_Time_slow epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Time_slow epoch`, 
  `median_RMSSD_Magnitude_fast epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_fast epoch`, 
  `median_RMSSD_Magnitude_medium epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_medium epoch`, 
  `median_RMSSD_Magnitude_slow epoch` = EVA_Patients_5min_RMSSD_joined$`median_RMSSD_Magnitude_slow epoch`, 
  `median_RMSSD_Time_total_epoch` = EVA_Patients_5min_RMSSD_joined$median_RMSSD_Time_total_epoch,
  `median_RMSSD_Magnitude_total_epoch` = EVA_Patients_5min_RMSSD_joined$median_RMSSD_Magnitude_total_epoch,
  `mcass_brsa` = EVA_Patients_5min_RMSSD_joined$`mcass_brsa`, 
  `mcass_brsv` = EVA_Patients_5min_RMSSD_joined$`mcass_brsv`, 
  `mcass_adrenergic_index` = EVA_Patients_5min_RMSSD_joined$`mcass_adrenergic_index`, 
  `mcass_sudomotor_index` = EVA_Patients_5min_RMSSD_joined$`mcass_sudomotor_index`, 
  `mcass_cardiovagal_index` = EVA_Patients_5min_RMSSD_joined$`mcass_cardiovagal_index`, 
  `mcass_total` = EVA_Patients_5min_RMSSD_joined$`mcass_total`
  ), y_vars, x_vars_new, method = "spearman", adjust = "none")

# long data for plotting
df_cor <- melt(sub_cor, varnames = c("Y","X"), value.name = "r")
df_p   <- melt(p_sub,   varnames = c("Y","X"), value.name = "p")

# flag significance (e.g., alpha = 0.05)
alpha <- 0.05
df_plot <- merge(df_cor, df_p, by = c("X","Y"))
df_plot$sig <- !is.na(df_plot$p) & df_plot$p < alpha

# --- 3) Plot A: correlation heatmap with numbers & significance marks ---
p_corr <- ggplot(df_plot, aes(X, Y, fill = r)) +
  geom_tile(color = "white") +
  geom_text(aes(label = sprintf("%.2f", r)), size = 3) +
  # add a star for significant cells
  # geom_text(data = subset(df_plot, sig), label = "*", vjust = -0.9, size = 4) +
  scale_fill_gradient2(low = "red", mid = "white", high = "blue",
                       midpoint = 0, limits = c(-1, 1), name = "Correlation") +
  coord_fixed() +
  theme_minimal(base_size = 12) +
  theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1),
        panel.grid = element_blank())

p_corr

# --- 4) Plot B: p-value heatmap with numbers (0–1 scale) ---
p_pval <- ggplot(df_plot, aes(X, Y, fill = p)) +
  geom_tile(color = "white") +
  geom_text(aes(label = ifelse(is.na(p), "", sprintf("%.3f", p))), size = 3) +
  scale_fill_gradient2(low = "red", high = "yellow",
                       limits = c(0, 1), name = "p-value") +
  coord_fixed() +
  theme_minimal(base_size = 12) +
  theme(axis.text.x = element_text(angle = 45, hjust = 1, vjust = 1),
        panel.grid = element_blank())

p_pval

# write.csv(EVA_Controls_5min, "Final_EVA_Controls_5min.csv")
# write.csv(EVA_Patients_5min, "Final_EVA_Patients_5min.csv")
# write.csv(Control_Patients_5min, "Final_Control_Patients_5min.csv")
# write.csv(EVA_Controls_5min_RMSSD_joined, "Final_EVA_Controls_5min_joined.csv")
# write.csv(EVA_Patients_5min_RMSSD_joined, "Final_EVA_Patients_5min_RMSSD_joined.csv")
# write.csv(Control_Patients_5min_RMSSD, "Final_Control_Patients_5min_RMSSD.csv")

6 Modification Analysis

Question: If the relationship between “number_of_contractions_over26.7_per_hour” and “mcass_brsa” varies depending on the value of “mcass_total” and/or “mcass_cardiovagal_index”?

6.1 mcass_total

Since mcass_total and mcass_brsa are correlated with each other, having multicollinearity issue, the independence assumption is violated.

6.2 mcass_cardiovagal_index

6.2.1 Poisson Regression

model <- glm(total_number_of_contractions_over26.7  ~ mcass_brsa, offset = log(total_no_missing_time), family = poisson(link = "log"), data = EVA_Patients_5min_general_excluding_normal)
sum(residuals(model, type = "pearson")^2)/df.residual(model)

## [1] 64.4671

Since Pearson Dispersion is much larger than 1, we should use Negative Binomial, instead of Poisson Regression.

6.2.2 Negative Binomial

6.2.2.1 Unadjusted Model

library(MASS)
summary(glm.nb(total_number_of_contractions_over26.7 ~ mcass_brsa + offset(log(total_no_missing_time)),  data = EVA_Patients_5min_general_excluding_normal))

## 
## Call:
## glm.nb(formula = total_number_of_contractions_over26.7 ~ mcass_brsa + 
##     offset(log(total_no_missing_time)), data = EVA_Patients_5min_general_excluding_normal, 
##     init.theta = 2.27066766, link = log)
## 
## Coefficients:
##             Estimate Std. Error z value Pr(>|z|)    
## (Intercept)  4.38710    0.22774   19.26  < 2e-16 ***
## mcass_brsa  -0.04152    0.01462   -2.84  0.00451 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for Negative Binomial(2.2707) family taken to be 1)
## 
##     Null deviance: 52.365  on 39  degrees of freedom
## Residual deviance: 42.968  on 38  degrees of freedom
## AIC: 474.64
## 
## Number of Fisher Scoring iterations: 1
## 
## 
##               Theta:  2.271 
##           Std. Err.:  0.488 
## 
##  2 x log-likelihood:  -468.636

# summary(glm.nb(total_number_of_contractions_over26.7 ~ mcass_cardiovagal_index + offset(log(total_no_missing_time)),  data = EVA_Patients_5min_general_excluding_normal))

The incidence rate ratio (IRR) is 0.9593321

For each 1-unit increase in mcass_brsa, the expected total_number_of_contractions_over26.7 rate per hour is mulltiplied by ~0.9593321, indicating a 4.07% decrease in the rate.

At mcass_brsa = x1, the rate is 10 contractions over 26.7 per hour, then at mcass_brsa = x1+1, the rate becomes 9.6 contractions over 26.7 per hour.

6.2.2.2 Confounding model

summary(glm.nb(total_number_of_contractions_over26.7 ~ mcass_brsa + mcass_cardiovagal_index + offset(log(total_no_missing_time)),  data = EVA_Patients_5min_general_excluding_normal))

## 
## Call:
## glm.nb(formula = total_number_of_contractions_over26.7 ~ mcass_brsa + 
##     mcass_cardiovagal_index + offset(log(total_no_missing_time)), 
##     data = EVA_Patients_5min_general_excluding_normal, init.theta = 2.675214968, 
##     link = log)
## 
## Coefficients:
##                         Estimate Std. Error z value Pr(>|z|)    
## (Intercept)              4.12724    0.23302  17.712  < 2e-16 ***
## mcass_brsa              -0.04440    0.01350  -3.288  0.00101 ** 
## mcass_cardiovagal_index  0.09310    0.03165   2.942  0.00326 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for Negative Binomial(2.6752) family taken to be 1)
## 
##     Null deviance: 61.399  on 39  degrees of freedom
## Residual deviance: 42.590  on 37  degrees of freedom
## AIC: 469.46
## 
## Number of Fisher Scoring iterations: 1
## 
## 
##               Theta:  2.675 
##           Std. Err.:  0.583 
## 
##  2 x log-likelihood:  -461.459

(exp(-0.04440) - exp(-0.04152))/exp(-0.04152)

## [1] -0.002875857

The percent change in the coefficient of mcass_brsa is much less than 10%, meaning that we conclude no material confounding by mcass_cardiovagal_index of the mcass_brsa - total_number_of_contractions_over26.7 per hour rate effect.

6.2.2.3 Effect Modification

If mcass_brsa’s effect on total_number_of_contractions_over26.7 per hour varies with mcass_cardiovagal_index, then there is effect modification.

summary(glm.nb(total_number_of_contractions_over26.7 ~ mcass_brsa * mcass_cardiovagal_index + offset(log(total_no_missing_time)),  data = EVA_Patients_5min_general_excluding_normal))

## 
## Call:
## glm.nb(formula = total_number_of_contractions_over26.7 ~ mcass_brsa * 
##     mcass_cardiovagal_index + offset(log(total_no_missing_time)), 
##     data = EVA_Patients_5min_general_excluding_normal, init.theta = 2.85876824, 
##     link = log)
## 
## Coefficients:
##                                     Estimate Std. Error z value Pr(>|z|)    
## (Intercept)                         4.523494   0.324119  13.956  < 2e-16 ***
## mcass_brsa                         -0.074773   0.021403  -3.494 0.000477 ***
## mcass_cardiovagal_index            -0.001703   0.057668  -0.030 0.976436    
## mcass_brsa:mcass_cardiovagal_index  0.007376   0.003682   2.003 0.045169 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for Negative Binomial(2.8588) family taken to be 1)
## 
##     Null deviance: 65.471  on 39  degrees of freedom
## Residual deviance: 42.415  on 36  degrees of freedom
## AIC: 468.56
## 
## Number of Fisher Scoring iterations: 1
## 
## 
##               Theta:  2.859 
##           Std. Err.:  0.626 
## 
##  2 x log-likelihood:  -458.560

(exp(-0.074773 + 0.007) - 1) * 100

## [1] -6.552743

Each increase of 1 unit of mcass_cardiovagal_index multiplies the per-unit effect of mcass_brsa by 1.007 -> 0.74% larger for Incidence Rate Ratio (change in the contractions over 26.7 per hour rate when X increases by 1).

Concrete example: The IRR for a +1 unit increase in mcass_brsa at a given mcass_cardiovagal_index is IRR = exp(-0.07 + 0.007*mcass_cardiovagal_index).

When mcass_cardiovagal_index = 0, increasing 1-unit for X reduces the rate of number of contractions over 26.7 per hour by 1 - exp(-0.074773 + 0.007376*0) = 7.2%.

When mcass_cardiovagal_index = 1, increasing 1-unit for X reduces the rate of number of contractions over 26.7 per hour by 1 - exp(-0.074773 + 0.007376*1) = 6.52%.

When mcass_cardiovagal_index = 10, increasing 1-unit for X does not change the rate of number of contractions over 26.7 per hour.

The modification effect is quite small.

7 Whether the key GI variables are influenced by the sex and age differences between the groups?

Question: Is the observed difference in the key GI variables between Jack’s data and EVA patient driven by sex/age imbalance, or does it persist after accounting for them?

In paper https://www.jacc.org/doi/10.1016/j.jacc.2016.10.060, the authors compared conventional covariate adjustment with several propensity score (PS) methods. Propensity score methods were not superior to the traditional covariate adjustment.

In line with prior methodological work (Elze et al., JACC 2017), which showed that propensity score methods were not superior to traditional covariate adjustment in observational analyses, it is believed that covariate adjustment in our regression model is sufficient.

7.1 Confounder Check

A variable is a confounder if it distorts the true relationship between the exposure and outcome. - That distortion is seen in how much the estimate (e.g., OR, β, HR) changes after adjustment — not whether the p-value crosses 0.05. Confounding is about bias in the effect estimate, not about statistical significance.

The percent change in the main effect (before vs. after adjustment) directly reflects how much the covariate alters your point estimate. It measures bias in the association, which is the essence of confounding.

If the estimate changes little (<10%), the variable doesn’t distort much.

If it changes meaningfully (≥10%), the variable materially biases your exposure–outcome relationship.

The percent change is calculated as \(Percent Change = \frac{|\text{Estimate}_{\text{adjusted}} - \text{Estimate}_{\text{crude}}|}{|\text{Estimate}_{\text{crude}}|} \times 100\%\).

When you look at whether your predictor remains statistically significant before and after adjustment, you’re assessing the stability and precision of its association after accounting for other variables — not confounding per se, but robustness of the effect.

Percent change in estimate - Whether another variable biases or distorts your predictor’s effect (confounding).

Change in p-value or significance - Whether your predictor’s effect remains statistically detectable after accounting for covariates (robustness / precision).

7.1.1 total_contractions_frequency - Negative Binomial - No confounding.

model <- glm(total_number_of_contractions  ~ source, offset = log(total_no_missing_time), family = poisson(link = "log"), data = all_data)
sum(residuals(model, type = "pearson")^2)/df.residual(model)

## [1] 161.6025

Since Pearson Dispersion is much larger than 1, we should use Negative Binomial, instead of Poisson Regression.

summary(glm.nb(total_number_of_contractions ~ source + offset(log(total_no_missing_time)), data = all_data))

## 
## Call:
## glm.nb(formula = total_number_of_contractions ~ source + offset(log(total_no_missing_time)), 
##     data = all_data, init.theta = 5.73464158, link = log)
## 
## Coefficients:
##                   Estimate Std. Error z value Pr(>|z|)    
## (Intercept)        5.74565    0.04973 115.541  < 2e-16 ***
## sourceJack's Data -0.32165    0.06532  -4.924 8.46e-07 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for Negative Binomial(5.7346) family taken to be 1)
## 
##     Null deviance: 198.45  on 168  degrees of freedom
## Residual deviance: 173.90  on 167  degrees of freedom
## AIC: 2445
## 
## Number of Fisher Scoring iterations: 1
## 
## 
##               Theta:  5.735 
##           Std. Err.:  0.612 
## 
##  2 x log-likelihood:  -2439.025

summary(glm.nb(total_number_of_contractions ~ source + Sex + offset(log(total_no_missing_time)), data = all_data))

## 
## Call:
## glm.nb(formula = total_number_of_contractions ~ source + Sex + 
##     offset(log(total_no_missing_time)), data = all_data, init.theta = 5.741597048, 
##     link = log)
## 
## Coefficients:
##                   Estimate Std. Error z value Pr(>|z|)    
## (Intercept)        5.76983    0.07105  81.204  < 2e-16 ***
## sourceJack's Data -0.33010    0.06723  -4.910 9.11e-07 ***
## SexMale           -0.03182    0.06802  -0.468     0.64    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for Negative Binomial(5.7416) family taken to be 1)
## 
##     Null deviance: 198.69  on 168  degrees of freedom
## Residual deviance: 173.90  on 166  degrees of freedom
## AIC: 2446.8
## 
## Number of Fisher Scoring iterations: 1
## 
## 
##               Theta:  5.742 
##           Std. Err.:  0.612 
## 
##  2 x log-likelihood:  -2438.809

(exp(summary(glm.nb(total_number_of_contractions ~ source + Sex  + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]) - exp(summary(glm.nb(total_number_of_contractions ~ source + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]))/exp(summary(glm.nb(total_number_of_contractions ~ source + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"])

## [1] -0.008416814

summary(glm.nb(total_number_of_contractions ~ source + Age  + offset(log(total_no_missing_time)), data = all_data))

## 
## Call:
## glm.nb(formula = total_number_of_contractions ~ source + Age + 
##     offset(log(total_no_missing_time)), data = all_data, init.theta = 5.910060164, 
##     link = log)
## 
## Coefficients:
##                    Estimate Std. Error z value Pr(>|z|)    
## (Intercept)        5.595557   0.117673  47.552  < 2e-16 ***
## sourceJack's Data -0.296837   0.066202  -4.484 7.33e-06 ***
## Age                0.002845   0.002019   1.409    0.159    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for Negative Binomial(5.9101) family taken to be 1)
## 
##     Null deviance: 198.82  on 167  degrees of freedom
## Residual deviance: 172.71  on 165  degrees of freedom
##   (1 observation deleted due to missingness)
## AIC: 2429.6
## 
## Number of Fisher Scoring iterations: 1
## 
## 
##               Theta:  5.910 
##           Std. Err.:  0.633 
## 
##  2 x log-likelihood:  -2421.605

(exp(summary(glm.nb(total_number_of_contractions ~ source + Age  + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]) - exp(summary(glm.nb(total_number_of_contractions ~ source + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]))/exp(summary(glm.nb(total_number_of_contractions ~ source + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"])

## [1] 0.02512134

summary(glm.nb(total_number_of_contractions ~ source + Sex + Age  + offset(log(total_no_missing_time)), data = all_data))

## 
## Call:
## glm.nb(formula = total_number_of_contractions ~ source + Sex + 
##     Age + offset(log(total_no_missing_time)), data = all_data, 
##     init.theta = 5.912453014, link = log)
## 
## Coefficients:
##                    Estimate Std. Error z value Pr(>|z|)    
## (Intercept)        5.610915   0.130169  43.105   <2e-16 ***
## sourceJack's Data -0.301879   0.068342  -4.417    1e-05 ***
## SexMale           -0.018154   0.067342  -0.270    0.787    
## Age                0.002815   0.002021   1.393    0.164    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for Negative Binomial(5.9125) family taken to be 1)
## 
##     Null deviance: 198.90  on 167  degrees of freedom
## Residual deviance: 172.71  on 164  degrees of freedom
##   (1 observation deleted due to missingness)
## AIC: 2431.5
## 
## Number of Fisher Scoring iterations: 1
## 
## 
##               Theta:  5.912 
##           Std. Err.:  0.633 
## 
##  2 x log-likelihood:  -2421.533

(exp(summary(glm.nb(total_number_of_contractions ~ source + Sex + Age  + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]) - exp(summary(glm.nb(total_number_of_contractions ~ source + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]))/exp(summary(glm.nb(total_number_of_contractions ~ source + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"])

## [1] 0.01996556

Since the percent change is less than 10%, there is no confounding.

7.1.2 number_of_contractions_over26.7_per_hour - Negative Binomial - No confounding.

model <- glm(total_number_of_contractions_over26.7  ~ source, offset = log(total_no_missing_time), family = poisson(link = "log"), data = all_data)
sum(residuals(model, type = "pearson")^2)/df.residual(model)

## [1] 68.88857

Since Pearson Dispersion is much larger than 1, we should use Negative Binomial, instead of Poisson Regression.

summary(glm.nb(total_number_of_contractions_over26.7 ~ source + offset(log(total_no_missing_time)), data = all_data))

## 
## Call:
## glm.nb(formula = total_number_of_contractions_over26.7 ~ source + 
##     offset(log(total_no_missing_time)), data = all_data, init.theta = 1.87209386, 
##     link = log)
## 
## Coefficients:
##                   Estimate Std. Error z value Pr(>|z|)    
## (Intercept)        3.83078    0.08739  43.834  < 2e-16 ***
## sourceJack's Data -0.51215    0.11491  -4.457 8.31e-06 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for Negative Binomial(1.8721) family taken to be 1)
## 
##     Null deviance: 203.34  on 168  degrees of freedom
## Residual deviance: 183.19  on 167  degrees of freedom
## AIC: 1903.6
## 
## Number of Fisher Scoring iterations: 1
## 
## 
##               Theta:  1.872 
##           Std. Err.:  0.193 
## 
##  2 x log-likelihood:  -1897.586

summary(glm.nb(total_number_of_contractions_over26.7 ~ source + Sex  + offset(log(total_no_missing_time)), data = all_data))

## 
## Call:
## glm.nb(formula = total_number_of_contractions_over26.7 ~ source + 
##     Sex + offset(log(total_no_missing_time)), data = all_data, 
##     init.theta = 1.897730656, link = log)
## 
## Coefficients:
##                   Estimate Std. Error z value Pr(>|z|)    
## (Intercept)         3.9645     0.1242  31.926  < 2e-16 ***
## sourceJack's Data  -0.5514     0.1176  -4.689 2.74e-06 ***
## SexMale            -0.1889     0.1190  -1.588    0.112    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for Negative Binomial(1.8977) family taken to be 1)
## 
##     Null deviance: 206.05  on 168  degrees of freedom
## Residual deviance: 183.05  on 166  degrees of freedom
## AIC: 1903
## 
## Number of Fisher Scoring iterations: 1
## 
## 
##               Theta:  1.898 
##           Std. Err.:  0.196 
## 
##  2 x log-likelihood:  -1895.020

(exp(summary(glm.nb(total_number_of_contractions_over26.7 ~ source + Sex  + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]) - exp(summary(glm.nb(total_number_of_contractions_over26.7 ~ source + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]))/exp(summary(glm.nb(total_number_of_contractions_over26.7 ~ source + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"])

## [1] -0.03845585

summary(glm.nb(total_number_of_contractions_over26.7 ~ source + Age  + offset(log(total_no_missing_time)), data = all_data))

## 
## Call:
## glm.nb(formula = total_number_of_contractions_over26.7 ~ source + 
##     Age + offset(log(total_no_missing_time)), data = all_data, 
##     init.theta = 1.973015419, link = log)
## 
## Coefficients:
##                    Estimate Std. Error z value Pr(>|z|)    
## (Intercept)        3.301979   0.204926  16.113  < 2e-16 ***
## sourceJack's Data -0.453430   0.115206  -3.936 8.29e-05 ***
## Age                0.010028   0.003516   2.852  0.00435 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for Negative Binomial(1.973) family taken to be 1)
## 
##     Null deviance: 209.27  on 167  degrees of freedom
## Residual deviance: 181.31  on 165  degrees of freedom
##   (1 observation deleted due to missingness)
## AIC: 1888
## 
## Number of Fisher Scoring iterations: 1
## 
## 
##               Theta:  1.973 
##           Std. Err.:  0.205 
## 
##  2 x log-likelihood:  -1879.990

(exp(summary(glm.nb(total_number_of_contractions_over26.7 ~ source + Age  + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]) - exp(summary(glm.nb(total_number_of_contractions_over26.7 ~ source + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]))/exp(summary(glm.nb(total_number_of_contractions_over26.7 ~ source + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"])

## [1] 0.06048331

summary(glm.nb(total_number_of_contractions_over26.7 ~ source + Sex + Age  + offset(log(total_no_missing_time)), data = all_data))

## 
## Call:
## glm.nb(formula = total_number_of_contractions_over26.7 ~ source + 
##     Sex + Age + offset(log(total_no_missing_time)), data = all_data, 
##     init.theta = 1.998034502, link = log)
## 
## Coefficients:
##                    Estimate Std. Error z value Pr(>|z|)    
## (Intercept)        3.430426   0.225307  15.226  < 2e-16 ***
## sourceJack's Data -0.490925   0.118250  -4.152  3.3e-05 ***
## SexMale           -0.177355   0.116560  -1.522  0.12812    
## Age                0.009971   0.003499   2.849  0.00438 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for Negative Binomial(1.998) family taken to be 1)
## 
##     Null deviance: 211.85  on 167  degrees of freedom
## Residual deviance: 181.17  on 164  degrees of freedom
##   (1 observation deleted due to missingness)
## AIC: 1887.6
## 
## Number of Fisher Scoring iterations: 1
## 
## 
##               Theta:  1.998 
##           Std. Err.:  0.208 
## 
##  2 x log-likelihood:  -1877.624

(exp(summary(glm.nb(total_number_of_contractions_over26.7 ~ source + Sex + Age  + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]) - exp(summary(glm.nb(total_number_of_contractions_over26.7 ~ source + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]))/exp(summary(glm.nb(total_number_of_contractions_over26.7 ~ source + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"])

## [1] 0.02145684

Since the percent change is less than 10%, there is no confounding.

7.1.3 number_of_total_epochs_frequency - Negative Binomial - No confounding.

model <- glm(number_of_total_epochs  ~ source, offset = log(total_no_missing_time), family = poisson(link = "log"), data = all_data)
sum(residuals(model, type = "pearson")^2)/df.residual(model)

## [1] 3.576141

Since Pearson Dispersion is much larger than 1, we should use Negative Binomial, instead of Poisson Regression.

summary(glm.nb(number_of_total_epochs ~ source + offset(log(total_no_missing_time)), data = all_data))

## 
## Call:
## glm.nb(formula = number_of_total_epochs ~ source + offset(log(total_no_missing_time)), 
##     data = all_data, init.theta = 4.242011971, link = log)
## 
## Coefficients:
##                   Estimate Std. Error z value Pr(>|z|)    
## (Intercept)        1.08434    0.07037  15.410   <2e-16 ***
## sourceJack's Data  0.21696    0.09057   2.395   0.0166 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for Negative Binomial(4.242) family taken to be 1)
## 
##     Null deviance: 184.66  on 168  degrees of freedom
## Residual deviance: 178.93  on 167  degrees of freedom
## AIC: 1074.8
## 
## Number of Fisher Scoring iterations: 1
## 
## 
##               Theta:  4.242 
##           Std. Err.:  0.665 
## 
##  2 x log-likelihood:  -1068.783

summary(glm.nb(number_of_total_epochs ~ source + Sex + offset(log(total_no_missing_time)), data = all_data))

## 
## Call:
## glm.nb(formula = number_of_total_epochs ~ source + Sex + offset(log(total_no_missing_time)), 
##     data = all_data, init.theta = 4.258986195, link = log)
## 
## Coefficients:
##                   Estimate Std. Error z value Pr(>|z|)    
## (Intercept)        1.12566    0.09913  11.355   <2e-16 ***
## sourceJack's Data  0.20210    0.09326   2.167   0.0302 *  
## SexMale           -0.05403    0.09323  -0.580   0.5622    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for Negative Binomial(4.259) family taken to be 1)
## 
##     Null deviance: 185.14  on 168  degrees of freedom
## Residual deviance: 179.06  on 166  degrees of freedom
## AIC: 1076.5
## 
## Number of Fisher Scoring iterations: 1
## 
## 
##               Theta:  4.259 
##           Std. Err.:  0.669 
## 
##  2 x log-likelihood:  -1068.452

(exp(summary(glm.nb(number_of_total_epochs ~ source + Sex + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]) - exp(summary(glm.nb(number_of_total_epochs ~ source + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]))/exp(exp(summary(glm.nb(number_of_total_epochs ~ source  + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]))

## [1] -0.005288966

summary(glm.nb(number_of_total_epochs ~ source + Age  + offset(log(total_no_missing_time)), data = all_data))

## 
## Call:
## glm.nb(formula = number_of_total_epochs ~ source + Age + offset(log(total_no_missing_time)), 
##     data = all_data, init.theta = 4.705615718, link = log)
## 
## Coefficients:
##                   Estimate Std. Error z value Pr(>|z|)    
## (Intercept)       0.675927   0.158405   4.267 1.98e-05 ***
## sourceJack's Data 0.287674   0.089868   3.201  0.00137 ** 
## Age               0.007539   0.002677   2.817  0.00485 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for Negative Binomial(4.7056) family taken to be 1)
## 
##     Null deviance: 191.29  on 167  degrees of freedom
## Residual deviance: 176.84  on 165  degrees of freedom
##   (1 observation deleted due to missingness)
## AIC: 1060.6
## 
## Number of Fisher Scoring iterations: 1
## 
## 
##               Theta:  4.706 
##           Std. Err.:  0.760 
## 
##  2 x log-likelihood:  -1052.603

(exp(summary(glm.nb(number_of_total_epochs ~ source + Age  + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]) - exp(summary(glm.nb(number_of_total_epochs ~ source + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]))/exp(exp(summary(glm.nb(number_of_total_epochs ~ source  + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]))

## [1] 0.02628221

summary(glm.nb(number_of_total_epochs ~ source + Sex + Age  + offset(log(total_no_missing_time)), data = all_data))

## 
## Call:
## glm.nb(formula = number_of_total_epochs ~ source + Sex + Age + 
##     offset(log(total_no_missing_time)), data = all_data, init.theta = 4.712933149, 
##     link = log)
## 
## Coefficients:
##                    Estimate Std. Error z value Pr(>|z|)    
## (Intercept)        0.699163   0.175250   3.990 6.62e-05 ***
## sourceJack's Data  0.279847   0.092886   3.013  0.00259 ** 
## SexMale           -0.026907   0.090414  -0.298  0.76601    
## Age                0.007489   0.002679   2.796  0.00518 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for Negative Binomial(4.7129) family taken to be 1)
## 
##     Null deviance: 191.47  on 167  degrees of freedom
## Residual deviance: 176.93  on 164  degrees of freedom
##   (1 observation deleted due to missingness)
## AIC: 1062.5
## 
## Number of Fisher Scoring iterations: 1
## 
## 
##               Theta:  4.713 
##           Std. Err.:  0.762 
## 
##  2 x log-likelihood:  -1052.516

(exp(summary(glm.nb(number_of_total_epochs ~ source + Sex + Age  + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]) - exp(summary(glm.nb(number_of_total_epochs ~ source + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]))/exp(exp(summary(glm.nb(number_of_total_epochs ~ source  + offset(log(total_no_missing_time)), data = all_data))$coefficients[2, "Estimate"]))

## [1] 0.02328061

Since the percent change is less than 10%, there is no confounding.

7.1.4 imputed_sum_of_amplitude - ANCOVA - No confounding.

summary(lm(imputed_sum_of_amplitude ~ source, data = all_data))

## 
## Call:
## lm(formula = imputed_sum_of_amplitude ~ source, data = all_data)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -125376  -42592  -12682   19110  763119 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)         187579      11253  16.669  < 2e-16 ***
## sourceJack's Data   -79402      14778  -5.373 2.57e-07 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 94820 on 167 degrees of freedom
## Multiple R-squared:  0.1474, Adjusted R-squared:  0.1423 
## F-statistic: 28.87 on 1 and 167 DF,  p-value: 2.566e-07

summary(lm(imputed_sum_of_amplitude ~ source + Sex, data = all_data))

## 
## Call:
## lm(formula = imputed_sum_of_amplitude ~ source + Sex, data = all_data)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -123975  -40760  -13004   18764  758996 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)         191702      16127  11.887  < 2e-16 ***
## sourceJack's Data   -80707      15259  -5.289 3.83e-07 ***
## SexMale              -5524      15437  -0.358    0.721    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 95070 on 166 degrees of freedom
## Multiple R-squared:  0.1481, Adjusted R-squared:  0.1378 
## F-statistic: 14.42 on 2 and 166 DF,  p-value: 1.676e-06

(summary(lm(imputed_sum_of_amplitude ~ source + Sex, data = all_data))$coefficients[2, "Estimate"] - summary(lm(imputed_sum_of_amplitude ~ source, data = all_data))$coefficients[2, "Estimate"])/summary(lm(imputed_sum_of_amplitude ~ source, data = all_data))$coefficients[2, "Estimate"]

## [1] 0.01643617

summary(lm(imputed_sum_of_amplitude ~ source + Age, data = all_data))

## 
## Call:
## lm(formula = imputed_sum_of_amplitude ~ source + Age, data = all_data)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -129893  -41744  -11400   20079  759955 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)       163675.3    27104.1   6.039 9.96e-09 ***
## sourceJack's Data -76239.1    15249.5  -4.999 1.46e-06 ***
## Age                  451.1      465.1   0.970    0.333    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 95110 on 165 degrees of freedom
##   (1 observation deleted due to missingness)
## Multiple R-squared:  0.1523, Adjusted R-squared:  0.142 
## F-statistic: 14.82 on 2 and 165 DF,  p-value: 1.204e-06

(summary(lm(imputed_sum_of_amplitude ~ source + Age, data = all_data))$coefficients[2, "Estimate"] - summary(lm(imputed_sum_of_amplitude ~ source, data = all_data))$coefficients[2, "Estimate"])/summary(lm(imputed_sum_of_amplitude ~ source, data = all_data))$coefficients[2, "Estimate"]

## [1] -0.03983097

summary(lm(imputed_sum_of_amplitude ~ source + Sex + Age, data = all_data))

## 
## Call:
## lm(formula = imputed_sum_of_amplitude ~ source + Sex + Age, data = all_data)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -128539  -42259  -10907   20427  756246 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)       167843.3    30068.8   5.582 9.66e-08 ***
## sourceJack's Data -77512.1    15787.8  -4.910 2.19e-06 ***
## SexMale            -5040.0    15556.0  -0.324    0.746    
## Age                  443.5      466.9   0.950    0.344    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 95370 on 164 degrees of freedom
##   (1 observation deleted due to missingness)
## Multiple R-squared:  0.1528, Adjusted R-squared:  0.1373 
## F-statistic: 9.861 on 3 and 164 DF,  p-value: 5.142e-06

(summary(lm(imputed_sum_of_amplitude ~ source + Sex + Age, data = all_data))$coefficients[2, "Estimate"] - summary(lm(imputed_sum_of_amplitude ~ source, data = all_data))$coefficients[2, "Estimate"])/summary(lm(imputed_sum_of_amplitude ~ source, data = all_data))$coefficients[2, "Estimate"]

## [1] -0.02379869

Since the percent change is less than 10%, there is no confounding.

7.1.5 Motility Index - ANCOVA - Confounding exists, but mild.

summary(lm(motility_index ~ source, data = all_data))

## 
## Call:
## lm(formula = motility_index ~ source, data = all_data)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -2.34781 -0.48502  0.08265  0.64249  2.15321 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)        18.3287     0.1064 172.283   <2e-16 ***
## sourceJack's Data  -0.3278     0.1397  -2.347   0.0201 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.8964 on 167 degrees of freedom
## Multiple R-squared:  0.03192,    Adjusted R-squared:  0.02612 
## F-statistic: 5.506 on 1 and 167 DF,  p-value: 0.02012

summary(lm(motility_index ~ source + Sex, data = all_data))

## 
## Call:
## lm(formula = motility_index ~ source + Sex, data = all_data)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -2.33189 -0.51706  0.07227  0.61399  2.18397 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)        18.3756     0.1524 120.542   <2e-16 ***
## sourceJack's Data  -0.3427     0.1442  -2.376   0.0187 *  
## SexMale            -0.0628     0.1459  -0.430   0.6675    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.8986 on 166 degrees of freedom
## Multiple R-squared:  0.033,  Adjusted R-squared:  0.02135 
## F-statistic: 2.832 on 2 and 166 DF,  p-value: 0.06173

(summary(lm(motility_index ~ source + Sex, data = all_data))$coefficients[2, "Estimate"]  - summary(lm(motility_index ~ source, data = all_data))$coefficients[2, "Estimate"])/summary(lm(motility_index ~ source, data = all_data))$coefficients[2, "Estimate"]

## [1] 0.04526403

summary(lm(motility_index ~ source + Age, data = all_data))

## 
## Call:
## lm(formula = motility_index ~ source + Age, data = all_data)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -2.29114 -0.51372  0.07746  0.63739  2.14190 
## 
## Coefficients:
##                    Estimate Std. Error t value Pr(>|t|)    
## (Intercept)       18.128285   0.253687  71.459   <2e-16 ***
## sourceJack's Data -0.282527   0.142731  -1.979   0.0494 *  
## Age                0.003782   0.004353   0.869   0.3862    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.8902 on 165 degrees of freedom
##   (1 observation deleted due to missingness)
## Multiple R-squared:  0.03364,    Adjusted R-squared:  0.02192 
## F-statistic: 2.872 on 2 and 165 DF,  p-value: 0.05944

(summary(lm(motility_index ~ source + Age, data = all_data))$coefficients[2, "Estimate"]  - summary(lm(motility_index ~ source, data = all_data))$coefficients[2, "Estimate"])/summary(lm(motility_index ~ source, data = all_data))$coefficients[2, "Estimate"]

## [1] -0.1381755

summary(lm(motility_index ~ source + Sex + Age, data = all_data))

## 
## Call:
## lm(formula = motility_index ~ source + Sex + Age, data = all_data)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -2.28308 -0.50494  0.07178  0.63631  2.15909 
## 
## Coefficients:
##                    Estimate Std. Error t value Pr(>|t|)    
## (Intercept)       18.157154   0.281476  64.507   <2e-16 ***
## sourceJack's Data -0.291344   0.147791  -1.971   0.0504 .  
## SexMale           -0.034909   0.145621  -0.240   0.8108    
## Age                0.003729   0.004371   0.853   0.3949    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.8928 on 164 degrees of freedom
##   (1 observation deleted due to missingness)
## Multiple R-squared:  0.03398,    Adjusted R-squared:  0.0163 
## F-statistic: 1.923 on 3 and 164 DF,  p-value: 0.1279

(summary(lm(motility_index ~ source + Sex + Age, data = all_data))$coefficients[2, "Estimate"]  - summary(lm(motility_index ~ source, data = all_data))$coefficients[2, "Estimate"])/summary(lm(motility_index ~ source, data = all_data))$coefficients[2, "Estimate"]

## [1] -0.111279

Adjusting for Sex, the percent change is less than 10%, indicating that Sex is not a confounder.

Adjusting for Age, the percent change is a little bit larger than 10%, indicating that Age is a mild confounder.

Adjusting for Sex and Age, the percent change is a little bit larger than 10%, indicating that Sex and Age are mild confounders jointly.

7.1.6 median_RMSSD_Time_fast epoch - ANCOVA - Confounding exists.

all_data_RMSSD <- all_data_RMSSD %>%
  right_join(all_data %>% dplyr::select(combined_ID, Age, Sex))

summary(lm(`median_RMSSD_Time_fast epoch` ~ source, data = all_data_RMSSD))

## 
## Call:
## lm(formula = `median_RMSSD_Time_fast epoch` ~ source, data = all_data_RMSSD)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -11.596  -3.928  -1.055   3.149  18.654 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)        53.1357     0.6817  77.943  < 2e-16 ***
## sourceJack's Data  -2.6704     0.8979  -2.974  0.00346 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 5.324 on 142 degrees of freedom
##   (25 observations deleted due to missingness)
## Multiple R-squared:  0.05863,    Adjusted R-squared:  0.052 
## F-statistic: 8.844 on 1 and 142 DF,  p-value: 0.003456

summary(lm(`median_RMSSD_Time_fast epoch` ~ source + Sex, data = all_data_RMSSD))

## 
## Call:
## lm(formula = `median_RMSSD_Time_fast epoch` ~ source + Sex, data = all_data_RMSSD)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -11.474  -3.858  -1.061   3.357  18.776 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)        53.5081     0.9864  54.246  < 2e-16 ***
## sourceJack's Data  -2.8048     0.9361  -2.996  0.00323 ** 
## SexMale            -0.4937     0.9431  -0.523  0.60146    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 5.338 on 141 degrees of freedom
##   (25 observations deleted due to missingness)
## Multiple R-squared:  0.06046,    Adjusted R-squared:  0.04713 
## F-statistic: 4.537 on 2 and 141 DF,  p-value: 0.01232

(summary(lm(`median_RMSSD_Time_fast epoch` ~ source + Sex, data = all_data_RMSSD))$coefficients[2,"Estimate"] - summary(lm(`median_RMSSD_Time_fast epoch` ~ source, data = all_data_RMSSD))$coefficients[2,"Estimate"])/summary(lm(`median_RMSSD_Time_fast epoch` ~ source, data = all_data_RMSSD))$coefficients[2, "Estimate"]

## [1] 0.05031982

summary(lm(`median_RMSSD_Time_fast epoch` ~ source + Age, data = all_data_RMSSD))

## 
## Call:
## lm(formula = `median_RMSSD_Time_fast epoch` ~ source + Age, data = all_data_RMSSD)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -11.1641  -3.4276  -0.4935   3.0769  18.0773 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)       56.70760    1.59210  35.618  < 2e-16 ***
## sourceJack's Data -3.16919    0.90496  -3.502 0.000619 ***
## Age               -0.06786    0.02744  -2.473 0.014588 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 5.231 on 141 degrees of freedom
##   (25 observations deleted due to missingness)
## Multiple R-squared:  0.09776,    Adjusted R-squared:  0.08497 
## F-statistic: 7.639 on 2 and 141 DF,  p-value: 0.000708

(summary(lm(`median_RMSSD_Time_fast epoch` ~ source + Age, data = all_data_RMSSD))$coefficients[2,"Estimate"] - summary(lm(`median_RMSSD_Time_fast epoch` ~ source, data = all_data_RMSSD))$coefficients[2,"Estimate"])/summary(lm(`median_RMSSD_Time_fast epoch` ~ source, data = all_data_RMSSD))$coefficients[2, "Estimate"]

## [1] 0.186767

summary(lm(`median_RMSSD_Time_fast epoch` ~ source + Sex + Age, data = all_data_RMSSD))

## 
## Call:
## lm(formula = `median_RMSSD_Time_fast epoch` ~ source + Sex + 
##     Age, data = all_data_RMSSD)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -11.0234  -3.5671  -0.6307   3.1170  17.8599 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)       57.15485    1.75910  32.491  < 2e-16 ***
## sourceJack's Data -3.32505    0.94299  -3.526 0.000571 ***
## SexMale           -0.55979    0.92670  -0.604 0.546774    
## Age               -0.06833    0.02751  -2.484 0.014182 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 5.243 on 140 degrees of freedom
##   (25 observations deleted due to missingness)
## Multiple R-squared:  0.1001, Adjusted R-squared:  0.08083 
## F-statistic: 5.192 on 3 and 140 DF,  p-value: 0.001977

(summary(lm(`median_RMSSD_Time_fast epoch` ~ source + Sex + Age, data = all_data_RMSSD))$coefficients[2,"Estimate"] - summary(lm(`median_RMSSD_Time_fast epoch` ~ source, data = all_data_RMSSD))$coefficients[2,"Estimate"])/summary(lm(`median_RMSSD_Time_fast epoch` ~ source, data = all_data_RMSSD))$coefficients[2, "Estimate"]

## [1] 0.2451344

Adjusting for Sex, the percent change is less than 10%, indicating that Sex is not a confounder.

Adjusting for Age, the percent change is larger than 10%, indicating that Age is a confounder.

Adjusting for Sex and Age, the percent change is larger than 10%, indicating that Sex and Age are confounders jointly.

7.1.7 median_RMSSD_Time_medium epoch - ANCOVA - Confounding exists, but mild.

summary(lm(`median_RMSSD_Time_medium epoch` ~ source, data = all_data_RMSSD))

## 
## Call:
## lm(formula = `median_RMSSD_Time_medium epoch` ~ source, data = all_data_RMSSD)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -16.652  -5.543  -1.468   2.962  34.213 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)         67.132      1.142  58.775   <2e-16 ***
## sourceJack's Data   -3.714      1.475  -2.519   0.0129 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 8.699 on 143 degrees of freedom
##   (24 observations deleted due to missingness)
## Multiple R-squared:  0.04248,    Adjusted R-squared:  0.03578 
## F-statistic: 6.343 on 1 and 143 DF,  p-value: 0.01288

summary(lm(`median_RMSSD_Time_medium epoch` ~ source + Sex, data = all_data_RMSSD))

## 
## Call:
## lm(formula = `median_RMSSD_Time_medium epoch` ~ source + Sex, 
##     data = all_data_RMSSD)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -16.874  -5.493  -1.413   2.527  33.991 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)        66.4941     1.6069  41.380   <2e-16 ***
## sourceJack's Data  -3.5012     1.5251  -2.296   0.0232 *  
## SexMale             0.8602     1.5209   0.566   0.5726    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 8.719 on 142 degrees of freedom
##   (24 observations deleted due to missingness)
## Multiple R-squared:  0.04463,    Adjusted R-squared:  0.03117 
## F-statistic: 3.317 on 2 and 142 DF,  p-value: 0.03911

(summary(lm(`median_RMSSD_Time_medium epoch` ~ source + Sex, data = all_data_RMSSD %>% filter(source %in% c("EVA Data", "Jack's Data"))))$coefficients[2,"Estimate"] - summary(lm(`median_RMSSD_Time_medium epoch` ~ source, data = all_data_RMSSD %>% filter(source %in% c("EVA Data", "Jack's Data"))))$coefficients[2,"Estimate"])/summary(lm(`median_RMSSD_Time_medium epoch` ~ source, data = all_data_RMSSD %>% filter(source %in% c("EVA Data", "Jack's Data"))))$coefficients[2, "Estimate"]

## [1] -0.05724126

summary(lm(`median_RMSSD_Time_medium epoch` ~ source + Age, data = all_data_RMSSD))

## 
## Call:
## lm(formula = `median_RMSSD_Time_medium epoch` ~ source + Age, 
##     data = all_data_RMSSD)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -16.687  -5.989  -1.408   2.881  33.958 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)       65.43632    2.70500  24.191   <2e-16 ***
## sourceJack's Data -3.46153    1.52161  -2.275   0.0244 *  
## Age                0.03147    0.04549   0.692   0.4902    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 8.715 on 142 degrees of freedom
##   (24 observations deleted due to missingness)
## Multiple R-squared:  0.04569,    Adjusted R-squared:  0.03225 
## F-statistic: 3.399 on 2 and 142 DF,  p-value: 0.03613

(summary(lm(`median_RMSSD_Time_medium epoch` ~ source + Age, data = all_data_RMSSD %>% filter(source %in% c("EVA Data", "Jack's Data"))))$coefficients[2,"Estimate"] - summary(lm(`median_RMSSD_Time_medium epoch` ~ source, data = all_data_RMSSD %>% filter(source %in% c("EVA Data", "Jack's Data"))))$coefficients[2,"Estimate"])/summary(lm(`median_RMSSD_Time_medium epoch` ~ source, data = all_data_RMSSD %>% filter(source %in% c("EVA Data", "Jack's Data"))))$coefficients[2, "Estimate"]

## [1] -0.06793226

summary(lm(`median_RMSSD_Time_medium epoch` ~ source + Sex + Age, data = all_data_RMSSD))

## 
## Call:
## lm(formula = `median_RMSSD_Time_medium epoch` ~ source + Sex + 
##     Age, data = all_data_RMSSD)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -16.944  -5.606  -1.559   2.703  33.680 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)       64.54611    3.04693  21.184   <2e-16 ***
## sourceJack's Data -3.19503    1.58066  -2.021   0.0451 *  
## SexMale            0.97979    1.53148   0.640   0.5234    
## Age                0.03451    0.04583   0.753   0.4528    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 8.733 on 141 degrees of freedom
##   (24 observations deleted due to missingness)
## Multiple R-squared:  0.04845,    Adjusted R-squared:  0.02821 
## F-statistic: 2.393 on 3 and 141 DF,  p-value: 0.07102

(summary(lm(`median_RMSSD_Time_medium epoch` ~ source + Sex + Age, data = all_data_RMSSD %>% filter(source %in% c("EVA Data", "Jack's Data"))))$coefficients[2,"Estimate"] - summary(lm(`median_RMSSD_Time_medium epoch` ~ source, data = all_data_RMSSD %>% filter(source %in% c("EVA Data", "Jack's Data"))))$coefficients[2,"Estimate"])/summary(lm(`median_RMSSD_Time_medium epoch` ~ source, data = all_data_RMSSD %>% filter(source %in% c("EVA Data", "Jack's Data"))))$coefficients[2, "Estimate"]

## [1] -0.1396935

Adjusting for Sex, the percent change is less than 10%, indicating that Sex is not a confounder.

Adjusting for Age, the percent change is larger than 10%, indicating that Age is a confounder.

Adjusting for Sex and Age, the percent change is larger than 10%, indicating that Sex and Age are mild confounders jointly.

This means that either variable alone is a confounder, but together they confound jointly, but mild.

7.1.8 median_RMSSD_Time_slow epoch - ANCOVA - No confounding.

summary(lm(`median_RMSSD_Time_slow epoch` ~ source, data = all_data_RMSSD))

## 
## Call:
## lm(formula = `median_RMSSD_Time_slow epoch` ~ source, data = all_data_RMSSD)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -26.258  -7.843  -1.263   5.857  54.354 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)         83.876      1.484  56.535  < 2e-16 ***
## sourceJack's Data   -5.417      1.909  -2.838  0.00518 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 11.4 on 147 degrees of freedom
##   (20 observations deleted due to missingness)
## Multiple R-squared:  0.05194,    Adjusted R-squared:  0.04549 
## F-statistic: 8.054 on 1 and 147 DF,  p-value: 0.005184

summary(lm(`median_RMSSD_Time_slow epoch` ~ source + Sex, data = all_data_RMSSD))

## 
## Call:
## lm(formula = `median_RMSSD_Time_slow epoch` ~ source + Sex, data = all_data_RMSSD)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -26.439  -7.741  -1.271   5.959  54.271 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)        83.5817     2.1573  38.744  < 2e-16 ***
## sourceJack's Data  -5.3202     1.9834  -2.682  0.00815 ** 
## SexMale             0.3772     2.0027   0.188  0.85087    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 11.43 on 146 degrees of freedom
##   (20 observations deleted due to missingness)
## Multiple R-squared:  0.05217,    Adjusted R-squared:  0.03919 
## F-statistic: 4.018 on 2 and 146 DF,  p-value: 0.02001

(summary(lm(`median_RMSSD_Time_slow epoch` ~ source + Sex, data = all_data_RMSSD))$coefficients[2,"Estimate"] - summary(lm(`median_RMSSD_Time_slow epoch` ~ source, data = all_data_RMSSD))$coefficients[2,"Estimate"])/summary(lm(`median_RMSSD_Time_slow epoch` ~ source, data = all_data_RMSSD))$coefficients[2,"Estimate"]

## [1] -0.01792457

summary(lm(`median_RMSSD_Time_slow epoch` ~ source + Age, data = all_data_RMSSD))

## 
## Call:
## lm(formula = `median_RMSSD_Time_slow epoch` ~ source + Age, data = all_data_RMSSD)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -24.769  -7.744  -1.401   6.113  53.276 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)       87.69717    3.38215  25.929  < 2e-16 ***
## sourceJack's Data -5.67444    1.93412  -2.934  0.00389 ** 
## Age               -0.07219    0.05754  -1.255  0.21162    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 11.29 on 145 degrees of freedom
##   (21 observations deleted due to missingness)
## Multiple R-squared:  0.05895,    Adjusted R-squared:  0.04597 
## F-statistic: 4.542 on 2 and 145 DF,  p-value: 0.01222

(summary(lm(`median_RMSSD_Time_slow epoch` ~ source + Age, data = all_data_RMSSD))$coefficients[2,"Estimate"] - summary(lm(`median_RMSSD_Time_slow epoch` ~ source, data = all_data_RMSSD))$coefficients[2,"Estimate"])/summary(lm(`median_RMSSD_Time_slow epoch` ~ source, data = all_data_RMSSD))$coefficients[2,"Estimate"]

## [1] 0.04746281

summary(lm(`median_RMSSD_Time_slow epoch` ~ source + Sex + Age, data = all_data_RMSSD))

## 
## Call:
## lm(formula = `median_RMSSD_Time_slow epoch` ~ source + Sex + 
##     Age, data = all_data_RMSSD)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -25.053  -7.892  -1.287   6.335  53.168 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)       87.22760    3.80111  22.948  < 2e-16 ***
## sourceJack's Data -5.52536    2.01514  -2.742  0.00688 ** 
## SexMale            0.54657    1.99447   0.274  0.78445    
## Age               -0.07137    0.05780  -1.235  0.21893    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 11.33 on 144 degrees of freedom
##   (21 observations deleted due to missingness)
## Multiple R-squared:  0.05944,    Adjusted R-squared:  0.03984 
## F-statistic: 3.033 on 3 and 144 DF,  p-value: 0.03129

(summary(lm(`median_RMSSD_Time_slow epoch` ~ source + Sex + Age, data = all_data_RMSSD))$coefficients[2,"Estimate"] - summary(lm(`median_RMSSD_Time_slow epoch` ~ source, data = all_data_RMSSD))$coefficients[2,"Estimate"])/summary(lm(`median_RMSSD_Time_slow epoch` ~ source, data = all_data_RMSSD))$coefficients[2,"Estimate"]

## [1] 0.01994353

Since the percent change is less than 10%, there is no confounding.

7.1.9 median_RMSSD_Magnitude_total_epoch - ANCOVA - Confounding exists.

summary(lm(median_RMSSD_Magnitude_total_epoch ~ source, data = all_data_RMSSD))

## 
## Call:
## lm(formula = median_RMSSD_Magnitude_total_epoch ~ source, data = all_data_RMSSD)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -37.822 -14.598  -4.507   8.362  84.012 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)         59.947      2.699  22.208   <2e-16 ***
## sourceJack's Data   -7.930      3.545  -2.237   0.0266 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 22.74 on 167 degrees of freedom
## Multiple R-squared:  0.02909,    Adjusted R-squared:  0.02328 
## F-statistic: 5.004 on 1 and 167 DF,  p-value: 0.02661

summary(lm(median_RMSSD_Magnitude_total_epoch ~ source + Sex, data = all_data_RMSSD))

## 
## Call:
## lm(formula = median_RMSSD_Magnitude_total_epoch ~ source + Sex, 
##     data = all_data_RMSSD)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -37.766 -14.559  -4.527   8.401  83.975 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)       59.89060    3.86997  15.476   <2e-16 ***
## sourceJack's Data -7.91177    3.66154  -2.161   0.0321 *  
## SexMale            0.07598    3.70438   0.021   0.9837    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 22.81 on 166 degrees of freedom
## Multiple R-squared:  0.0291, Adjusted R-squared:  0.0174 
## F-statistic: 2.487 on 2 and 166 DF,  p-value: 0.08622

((summary(lm(median_RMSSD_Magnitude_total_epoch ~ source + Sex, data = all_data_RMSSD))$coefficients[2,"Estimate"] - summary(lm(median_RMSSD_Magnitude_total_epoch ~ source, data = all_data_RMSSD))$coefficients[2,"Estimate"])/summary(lm(median_RMSSD_Magnitude_total_epoch ~ source, data = all_data_RMSSD))$coefficients[2,"Estimate"])

## [1] -0.00226397

summary(lm(median_RMSSD_Magnitude_total_epoch ~ source + Age, data = all_data_RMSSD))

## 
## Call:
## lm(formula = median_RMSSD_Magnitude_total_epoch ~ source + Age, 
##     data = all_data_RMSSD)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -38.717 -14.703  -4.146   9.554  76.520 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)        44.4576     6.3534   6.997 6.22e-11 ***
## sourceJack's Data  -5.4734     3.5746  -1.531  0.12763    
## Age                 0.2923     0.1090   2.682  0.00807 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 22.3 on 165 degrees of freedom
##   (1 observation deleted due to missingness)
## Multiple R-squared:  0.06782,    Adjusted R-squared:  0.05652 
## F-statistic: 6.002 on 2 and 165 DF,  p-value: 0.003047

((summary(lm(median_RMSSD_Magnitude_total_epoch ~ source + Age, data = all_data_RMSSD))$coefficients[2,"Estimate"] - summary(lm(median_RMSSD_Magnitude_total_epoch ~ source, data = all_data_RMSSD))$coefficients[2,"Estimate"])/summary(lm(median_RMSSD_Magnitude_total_epoch ~ source, data = all_data_RMSSD))$coefficients[2,"Estimate"])

## [1] -0.3097576

summary(lm(median_RMSSD_Magnitude_total_epoch ~ source + Sex + Age, data = all_data_RMSSD))

## 
## Call:
## lm(formula = median_RMSSD_Magnitude_total_epoch ~ source + Sex + 
##     Age, data = all_data_RMSSD)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -38.030 -14.437  -4.090   9.522  76.267 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)        43.6788     7.0492   6.196  4.5e-09 ***
## sourceJack's Data  -5.2355     3.7012  -1.415  0.15910    
## SexMale             0.9418     3.6469   0.258  0.79653    
## Age                 0.2938     0.1095   2.684  0.00803 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 22.36 on 164 degrees of freedom
##   (1 observation deleted due to missingness)
## Multiple R-squared:  0.0682, Adjusted R-squared:  0.05115 
## F-statistic: 4.001 on 3 and 164 DF,  p-value: 0.008809

((summary(lm(median_RMSSD_Magnitude_total_epoch ~ source + Sex + Age, data = all_data_RMSSD))$coefficients[2,"Estimate"] - summary(lm(median_RMSSD_Magnitude_total_epoch ~ source, data = all_data_RMSSD))$coefficients[2,"Estimate"])/summary(lm(median_RMSSD_Magnitude_total_epoch ~ source, data = all_data_RMSSD))$coefficients[2,"Estimate"])

## [1] -0.3397569

Adjusting for Sex, the percent change is less than 10%, indicating that Sex is not a confounder.

Adjusting for Age, the percent change is larger than 10%, indicating that Age is a confounder.

Adjusting for Sex and Age, the percent change is larger than 10%, indicating that Sex and Age are confounders jointly.

7.1.10 total_contraction_rate_over10 - Quasi Binomial - Confounder exists, but mild.

model <- glm(cbind(total_contraction_rate_over10 * total_minute_contraction, total_minute_contraction - total_contraction_rate_over10*total_minute_contraction) ~ source, family = binomial(), data = all_data)
sum(residuals(model, type = "pearson")^2)/df.residual(model)

## [1] 74.59594

Since Pearson Dispersion is much larger than 1, we should use Quasi-Binomial, instead of Binomial regression.

summary(glm(cbind(total_contraction_rate_over10 * total_minute_contraction, total_minute_contraction - total_contraction_rate_over10*total_minute_contraction) ~ source, family = quasibinomial(), data = all_data))

## 
## Call:
## glm(formula = cbind(total_contraction_rate_over10 * total_minute_contraction, 
##     total_minute_contraction - total_contraction_rate_over10 * 
##         total_minute_contraction) ~ source, family = quasibinomial(), 
##     data = all_data)
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)        -3.5965     0.2011 -17.888  < 2e-16 ***
## sourceJack's Data   1.2033     0.2929   4.108 6.23e-05 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for quasibinomial family taken to be 74.59594)
## 
##     Null deviance: 9086.6  on 168  degrees of freedom
## Residual deviance: 7897.1  on 167  degrees of freedom
## AIC: NA
## 
## Number of Fisher Scoring iterations: 7

summary(glm(cbind(total_contraction_rate_over10 * total_minute_contraction, total_minute_contraction - total_contraction_rate_over10*total_minute_contraction) ~ source + Sex, family = quasibinomial(), data = all_data))

## 
## Call:
## glm(formula = cbind(total_contraction_rate_over10 * total_minute_contraction, 
##     total_minute_contraction - total_contraction_rate_over10 * 
##         total_minute_contraction) ~ source + Sex, family = quasibinomial(), 
##     data = all_data)
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)        -4.1466     0.2395 -17.313  < 2e-16 ***
## sourceJack's Data   1.0192     0.2565   3.973 0.000106 ***
## SexMale             1.1674     0.2718   4.295 2.97e-05 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for quasibinomial family taken to be 55.22838)
## 
##     Null deviance: 9086.6  on 168  degrees of freedom
## Residual deviance: 6795.8  on 166  degrees of freedom
## AIC: NA
## 
## Number of Fisher Scoring iterations: 7

(summary(glm(cbind(total_contraction_rate_over10 * total_minute_contraction, total_minute_contraction - total_contraction_rate_over10*total_minute_contraction) ~ source + Sex, family = quasibinomial(), data = all_data))$coefficients[2,"Estimate"] - summary(glm(cbind(total_contraction_rate_over10 * total_minute_contraction, total_minute_contraction - total_contraction_rate_over10*total_minute_contraction) ~ source, family = quasibinomial(), data = all_data))$coefficients[2,"Estimate"])/summary(glm(cbind(total_contraction_rate_over10 * total_minute_contraction, total_minute_contraction - total_contraction_rate_over10*total_minute_contraction) ~ source, family = quasibinomial(), data = all_data))$coefficients[2,"Estimate"]

## [1] -0.1530151

summary(glm(cbind(total_contraction_rate_over10 * total_minute_contraction, total_minute_contraction - total_contraction_rate_over10*total_minute_contraction) ~ source + Age, family = quasibinomial(), data = all_data))

## 
## Call:
## glm(formula = cbind(total_contraction_rate_over10 * total_minute_contraction, 
##     total_minute_contraction - total_contraction_rate_over10 * 
##         total_minute_contraction) ~ source + Age, family = quasibinomial(), 
##     data = all_data)
## 
## Coefficients:
##                    Estimate Std. Error t value Pr(>|t|)    
## (Intercept)       -3.252373   0.613289  -5.303 3.61e-07 ***
## sourceJack's Data  1.143346   0.319005   3.584 0.000445 ***
## Age               -0.006221   0.010519  -0.591 0.555048    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for quasibinomial family taken to be 74.39515)
## 
##     Null deviance: 9068.6  on 167  degrees of freedom
## Residual deviance: 7832.0  on 165  degrees of freedom
##   (1 observation deleted due to missingness)
## AIC: NA
## 
## Number of Fisher Scoring iterations: 7

(summary(glm(cbind(total_contraction_rate_over10 * total_minute_contraction, total_minute_contraction - total_contraction_rate_over10*total_minute_contraction) ~ source + Age, family = quasibinomial(), data = all_data))$coefficients[2,"Estimate"] - summary(glm(cbind(total_contraction_rate_over10 * total_minute_contraction, total_minute_contraction - total_contraction_rate_over10*total_minute_contraction) ~ source, family = quasibinomial(), data = all_data))$coefficients[2,"Estimate"])/summary(glm(cbind(total_contraction_rate_over10 * total_minute_contraction, total_minute_contraction - total_contraction_rate_over10*total_minute_contraction) ~ source, family = quasibinomial(), data = all_data))$coefficients[2,"Estimate"]

## [1] -0.04980738

summary(glm(cbind(total_contraction_rate_over10 * total_minute_contraction, total_minute_contraction - total_contraction_rate_over10*total_minute_contraction) ~ source + Sex + Age, family = quasibinomial(), data = all_data))

## 
## Call:
## glm(formula = cbind(total_contraction_rate_over10 * total_minute_contraction, 
##     total_minute_contraction - total_contraction_rate_over10 * 
##         total_minute_contraction) ~ source + Sex + Age, family = quasibinomial(), 
##     data = all_data)
## 
## Coefficients:
##                    Estimate Std. Error t value Pr(>|t|)    
## (Intercept)       -4.376397   0.565896  -7.734 1.00e-12 ***
## sourceJack's Data  1.061534   0.264569   4.012 9.12e-05 ***
## SexMale            1.206556   0.281689   4.283 3.13e-05 ***
## Age                0.003921   0.008951   0.438    0.662    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for quasibinomial family taken to be 55.89491)
## 
##     Null deviance: 9068.6  on 167  degrees of freedom
## Residual deviance: 6726.3  on 164  degrees of freedom
##   (1 observation deleted due to missingness)
## AIC: NA
## 
## Number of Fisher Scoring iterations: 7

(summary(glm(cbind(total_contraction_rate_over10 * total_minute_contraction, total_minute_contraction - total_contraction_rate_over10*total_minute_contraction) ~ source + Sex + Age, family = quasibinomial(), data = all_data))$coefficients[2,"Estimate"] - summary(glm(cbind(total_contraction_rate_over10 * total_minute_contraction, total_minute_contraction - total_contraction_rate_over10*total_minute_contraction) ~ source, family = quasibinomial(), data = all_data))$coefficients[2,"Estimate"])/summary(glm(cbind(total_contraction_rate_over10 * total_minute_contraction, total_minute_contraction - total_contraction_rate_over10*total_minute_contraction) ~ source, family = quasibinomial(), data = all_data))$coefficients[2,"Estimate"]

## [1] -0.1177983

Adjusting for Sex, the percent change is larger than 10%, indicating that Sex is not a confounder.

Adjusting for Age, the percent change is less than 10%, indicating that Age is a confounder.

Adjusting for Sex and Age, the percent change is larger than 10%, indicating that Sex and Age are mild confounders jointly.

7.2 Balance the data

Plot_conditioning(all_data, c("Age", "Sex"), "source")

library(forcats)
mod <- glm(source_cat ~ Age + Sex, family = binomial(link = "logit"), data = all_data %>% ungroup() %>% mutate(source_cat = ifelse(source == "EVA Patient", 1, 0)) %>% dplyr::select(source_cat, Age, Sex) %>% na.omit())
summary(mod)

## 
## Call:
## glm(formula = source_cat ~ Age + Sex, family = binomial(link = "logit"), 
##     data = all_data %>% ungroup() %>% mutate(source_cat = ifelse(source == 
##         "EVA Patient", 1, 0)) %>% dplyr::select(source_cat, Age, 
##         Sex) %>% na.omit())
## 
## Coefficients:
##             Estimate Std. Error z value Pr(>|z|)    
## (Intercept) -2.55444    0.62271  -4.102 4.09e-05 ***
## Age          0.03119    0.01060   2.941  0.00327 ** 
## SexMale      1.11607    0.35148   3.175  0.00150 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for binomial family taken to be 1)
## 
##     Null deviance: 228.86  on 167  degrees of freedom
## Residual deviance: 209.39  on 165  degrees of freedom
## AIC: 215.39
## 
## Number of Fisher Scoring iterations: 4

# all_data %>%
#   ungroup() %>%
#   select(source, Age, Sex) %>%
#   na.omit() %>%
#   mutate(p = as.vector(predict(mod, type = "response"))) %>%
#   ggplot() +
#   geom_histogram(aes(x = p, fill = source), alpha = 0.3)

7.2.1 Propensity Score Matching

library(MatchIt)
df <- all_data %>%
  ungroup() %>%
  dplyr::select(source, Age, Sex, everything()) %>%
  tidyr::drop_na(source, Age, Sex) %>%
  mutate(
    # your propensity from a fitted model 'mod'
    p = as.vector(predict(mod, type = "response")),
    # make sure treatment is binary/factor with 2 levels
    source = as.factor(source)
  )

m.out <- matchit(
  source ~ Age + Sex,
  data        = df,
  method      = "nearest",
  estimand    = "ATT",
  distance    = df$p,      # use your prefit propensity scores
  discard     = "both",    # <-- auto-drop outside overlap
  caliper     = 0.2,
  std.caliper = TRUE,
  replace     = FALSE
)
summary(m.out)

## 
## Call:
## matchit(formula = source ~ Age + Sex, data = df, method = "nearest", 
##     distance = df$p, estimand = "ATT", discard = "both", replace = FALSE, 
##     caliper = 0.2, std.caliper = TRUE)
## 
## Summary of Balance for All Data:
##           Means Treated Means Control Std. Mean Diff. Var. Ratio eCDF Mean
## distance         0.3761        0.4862         -0.6631     1.3893    0.1996
## Age             45.5773       52.9859         -0.4055     2.3896    0.1516
## SexFemale        0.4948        0.2535          0.4827          .    0.2413
## SexMale          0.5052        0.7465         -0.4827          .    0.2413
##           eCDF Max
## distance    0.3922
## Age         0.3279
## SexFemale   0.2413
## SexMale     0.2413
## 
## Summary of Balance for Matched Data:
##           Means Treated Means Control Std. Mean Diff. Var. Ratio eCDF Mean
## distance         0.4518        0.4399          0.0713     0.8746    0.0272
## Age             50.6042       51.0417         -0.0239     1.7403    0.0618
## SexFemale        0.3125        0.3750         -0.1250          .    0.0625
## SexMale          0.6875        0.6250          0.1250          .    0.0625
##           eCDF Max Std. Pair Dist.
## distance    0.1250          0.0749
## Age         0.1458          0.4937
## SexFemale   0.0625          0.4584
## SexMale     0.0625          0.4584
## 
## Sample Sizes:
##           Control Treated
## All            71      97
## Matched        48      48
## Unmatched      23      38
## Discarded       0      11

library(cobalt)

# SMDs before (M.Raw) and after (M.Adj) matching
bt <- bal.tab(m.out, un = TRUE, s.d.denom = "pooled",
              m.threshold = 0.10,          # flag |SMD| > 0.10
              v.threshold = 2)             # optional: flag variance ratio outside [0.5, 2]

tab <- as.data.frame(bt$Balance)
# Most cobalt versions name the SMD columns as Diff.Un (pre) and Diff.Adj (post)
smds <- tab[, intersect(c("Diff.Un","Diff.Adj","M.Raw","M.Adj"), names(tab))]
round(smds, 3)                 # print nicely rounded

##          Diff.Un Diff.Adj
## distance  -0.715    0.077
## Age       -0.481   -0.028
## Sex_Male  -0.241    0.062

The standardized mean differences (SMD) of Age and Sex without propensity score matching are larger than 0.1, indicating imbalance. After propensity score matching, the SMDs are lower than 0.1, and the imbalance of Age and Sex has been solved.

# matched_all_data <- all_data %>% 
#   ungroup() %>% 
#   dplyr::select(source, Age, Sex, total_contractions_frequency) %>% 
#   na.omit() %>%
#   mutate(p = as.vector(predict(mod, type = "response"))) %>%
#   filter(p > 0.3 & p < 0.6)

matched_all_data <- match.data(m.out) %>% filter(weights > 0)